The Research Objective Study of the human genome demonstrates

Study of the human genome demonstrates that nearly all diseases have a genetic/genomic

component (“FAQ About Genetic Disorders,” 2012). Genetics and genomics differ in that

genetics focuses on the structure and function of a single gene, while genomics evaluates all of

the genes and their interactions within an organism (“World Health Organization [WHO]

definitions of genetics and genomics,” n.d.). Clinical applications of genetics and genomics

include pharmacogenomics, screening for genetic variation associated with increased risk of

disease, evaluating if individuals are carriers of genetic variants associated with diseases, and

providing patient counseling and education (“Applying an Implementation Science Approach to

Genomic Medicine,” n.d.). Clinical applications of genetics/genomics include not only

identifying individuals at risk for diseases, but also ameliorating risk through interventions such

as targeted screening and informed therapeutic decision-making.

Because genetics/genomics play a role in disease prevention and health promotion,

screening, diagnosis, treatment selection, and patient-education, all health care providers must

adopt the use of genetics and genomics into clinical practice to provide competent care.

However, despite its importance, evidence indicates genetics and genomics are not being fully

adopted into clinical nursing practice (Calzone, Jenkins, Culp, Bonham, & Badzek, 2013). There

are a number of reasons for this lack of adoption, including lack of knowledge, a lack of

understanding of the relevance, inadequate confidence to use this knowledge to adopt

genetics/genomics into nursing practice, as well as inadequate support from social system(s)

(Calzone, Jenkins, Culp, Caskey, & Badzek, 2014; Camak, 2016).

To increase adoption of genetics/genomics into nursing practice, nursing leaders from

clinical, academic, and research settings worked collaboratively to identify nurses’ roles and

responsibilities in genetics/genomics that are applicable for nurses in all clinical specialties,

roles, and practice settings (Jenkins & Calzone, 2007). This work led to the development of the

Genetics and Genomics Nursing Practice Survey (GGNPS). The GGNPS is an instrument

evaluating nurses’ competency/knowledge, confidence, attitudes/receptivity, and

decision/adoption of genetics genomics into nursing practice; in addition, the GGNPS evaluates

the effect of social systems on each of the previously mentioned variables (Calzone et al., 2012).

The GGNPS can be used both to promote and evaluate the goals of competency initiatives, such

as increasing adoption of genetics/genomics into nursing practice.

The GGNPS was developed from a previously validated instrument assessing the

adoption of genetics/genomics by family physicians (FPs). The GGNPS was revised to reflect

nursing practice and evaluated for test-retest reliability (Calzone et al., 2012, 2016). Although

the FP instrument was evaluated using structural equation modeling and content expert feedback,

the GGNPS has not undergone the same validity evaluation. Thus, it is currently not known if

the GGNPS accurately measures the above domains, in nursing practice. The purpose of this

study was to evaluate the GGNPS for face, content, and construct validity.

The Problem/Research Question

Is the Genetics and Genomics Nursing Practice Survey (GGNPS) a valid instrument to

evaluate registered nurses’ competency/knowledge, confidence, attitudes/receptivity and

decision/adoption of genetics and genomics into nursing practice, and the effect of social

systems?

Definitions

Variables are defined in quotes to provide exact definitions and prevent inadvertent

changes to the definitions of the terms by rephrasing.

Competency/knowledge. Competency was conceptually defined as individuals

providing safe patient care, in accordance with “responsibilities, professional standards,

education, and qualifications” (Axley, 2008, p. 221). Knowledge was conceptually defined as

occurring “when an individual (or other decision-making unit) is exposed to an innovation’s

existence and gains some understanding of how it functions.” Competency/knowledge was

operationally defined as self-reported knowledge of genetics and genomics, genetic risk and

family history assessment. GGNPS: Part 2, Questions 2.1, 2.2, 2.4; Part 4, Questions 1,

2, 3; Part 5, Questions 1, 2; Part 6, Questions 1, 2.

Attitude/receptivity. Attitude is “a relatively enduring organization of an individual’s

beliefs about an object that predisposes his or her actions” (Rogers, 2003, p. 174–175).

Receptivity was conceptually defined as forming a “favorable or unfavorable attitude towards an

innovation” (Rogers, 2003, p. 169). While knowledge is mostly 'knowing' or cognitive,

attitude/receptivity is mostly 'feeling' (Rogers, 2003). Attitude/receptivity was operationally

defined as GGNPS: Part 1, Questions, 1, 2, 3; Part 2, Questions 2, 3.

Decision/adoption. Decision was conceptually defined as activities that lead to a choice

of either adopting or rejecting the innovation; adoption was conceptually defined as use of an

innovation (Rogers, 2003). In the present study, decision/adoption was operationally defined as

self-reported collection and assessment of a family history, as well as self-reported facilitation of

referrals to genetic services. GGNPS: Part 3, Questions 2, 3, 4.

Confidence. “Level of certainty that knowledge about the innovation is accurate”

(Calzone et al., 2012, p. 12). Confidence was operationally defined as responses to confidence-

related questions on the GGNPS: Part 2, Question 1.

Social system. Social system was conceptually defined as the setting or environment

where the innovation was introduced such as the clinical site where nurses are employed

(Calzone et al., 2012). Social system was operationally defined as responses to the social

system-related questions on the GGNPS: Part 7, Questions 1.3, 1.4, 1.5, 1.6, and 1.7.

Delimitations

The present study was a secondary analysis of data collected at 23 Magnet® Recognition

Program hospitals and included a sample of 7,798 registered nurses. These nurses worked in 17

states, from all regions of the United States. Hospitals where the data was collected included a

rural hospital, three children’s hospitals, a Veteran’s Administration, a psychiatric hospital, and

one cancer center (Calzone et al., 2014).

The sample was large and from a variety of clinical sites, but all of the clinical locations

were Magnet® Recognition Program hospitals. A designation based on the American Nurses

Credentialing Center's Magnet Recognition Program, Magnet® hospitals are believed to provide

higher quality of nursing care; a study evaluating nursing care in Magnet® and non-Magnet®

hospitals reported increased patient teaching, more staffing resources, and better communication

in Magnet® hospitals (Kalisch & Xie, 2014). Due to these characteristics, Magnet® Recognition

Program hospitals have an increased capacity to innovate and support pilot programs for new

initiatives (Calzone et al., 2014). As a result, the researchers used purposive sample in selecting

Magnet® Recognition Program hospitals for their sample. The Magnet®

Recognition Program is based on national standards. Therefore, although the sample was only

from Magnet® hospitals, the selected hospitals are more likely to have increased consistency in

infrastructure, educational support, and resources.

Theoretical Rationale

Rogers’ (2003) Diffusion of Innovations (DOI) model provided a conceptual framework

for understanding the adoption of genetics and genomics into clinical nursing practice. Diffusion

of innovations is a process whereby innovations are “communicated through certain channels

over time among members of a social system” (p. 5). According to Rogers, diffusion of

innovation is a process or series of processes whereby an individual (or group) initially acquires

knowledge about an innovation, forms a favorable or unfavorable attitude about the innovation,

and then decides to either adopt or reject it. Rogers also asserted that personality variables affect

adoption; he reported that earlier adopters have a more positive attitude towards change and are

better able to cope with uncertainty and risk. Therefore, this study considered the direct effect of

confidence on attitudes/receptivity. The study also evaluated the indirect effect of confidence on

the outcome variable of adoption of innovations. Likewise, the social system in which the

innovation was introduced was proposed to have both direct and indirect effects. That is, the

social system had a direct influence on all three processes (knowledge, attitude, and adoption),

but also influenced both attitude and adoption indirectly through its direct effect on knowledge

and adoption indirectly through its direct effect on attitude.

Figure 1

Rogers’ DOI model adapted for this study.

Social Systems

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

Rogers’ DOI model also includes additional variables such as implementation and

confirmation (see Appendix A); however, this study only evaluated DOI variables measured in

the GGNPS. This did not include the variables implementation and confirmation. Rogers (2003)

defines adoption as full use of an innovation, while implementation is associated with some

uncertainty regarding the innovation and possible reinvention or attempts to modify the

innovation to increase usability. Confirmation is defined as an individual’s personal evaluation

of how they are using the innovation. Further studies evaluating if nurses have continued

uncertainty regarding genetics/genomics or attempt to increase usability by modifying how they

use it in their practice can be used to evaluate implementation and confirmation.

Although molecular scientists and health care researchers have studied genetics and genomics

extensively for a prolonged period of time, they fit the criteria for an innovation in clinical

practice settings because nurses lack knowledge of genetics/genomics and have been slow to

adopt this content into their clinical practice. In addition, the field of genetics and genomics is

continuously expanding with additional molecular discoveries and clinical applications.

Although sequencing of the human genome was completed in 2001, researchers

are still using this information to understand the roles of genes in biological functions, the

interaction between genes and the environment, and how the study of genetic variation can be

applied to health promotion and disease prevention (McAllister et al., 2017). An additional

example is the Precision Medicine Initiative, an effort to pool data from a cohort of one million

people to accelerate clinical research and inform patient care (Collins & Varmus, 2015).

Adoption was the ultimate outcome variable that was evaluated using the GGNPS and is

defined by Rogers (2003) as the full-scale use of an innovation. Rogers’ theory also attempts to

explain how innovations spread among social systems and what factors affect their diffusion.

Rogers stated individuals would adopt an innovation if they believed it has advantages or

usefulness that are not currently available. However, adoption is also affected by a number of

variables, such as the characteristics of the innovation, the adopters, and the system where

adoption is occurring.

GGNPS indicators of adoption of genetics/genomics included collection and evaluation

of a family history, facilitating referrals to specialists when required, and applying knowledge of

clinical genetics and genomics to provide competent and current patient care. These indicators of

adoption were pertinent because they applied to nurses of all levels of academic preparation in a

variety of clinical areas, in different specialties, and with variable access to technology.

Collection and evaluation of a family history, facilitation of referrals, and patient education are

pertinent to nurses caring for clients in in-patient and community settings. As well as nurses

working with patients across the lifespan, using a variety of documentation systems (paper or

electronic), and with varying access to sequencing and genetic technology. Adoption was a key

step because even if an individual has the knowledge to use an innovation, a positive attitude

towards it, and decided to use it -without adoption- the innovation cannot be integrated into

current practice. According to Rogers (2003), the decision to adopt an innovation is the

immediate precursor to adoption of an innovation.

Decision was the portion of the DOI model where individuals engaged in activities that

affected the choice of adopting or rejecting an innovation. However, the decision stage is

difficult to observe and measure because decisions are often internal thoughts (Rogers, 2003).

For this reason, Calzone and colleagues combined decision and adoption into one domain,

facilitating empirical measurement. In this study, decision was considered part of

decision/adoption and is not measured separately.

According to Rogers (2003), decision/adoption is directly affected by one’s

attitudes/receptivity towards the innovation. All innovations engender some degree of

uncertainty for the user, who may be unsure about its utility or may seek reinforcement from

others to form an opinion about an innovation. Positive attitudes/receptivity are critical to

adoption of the innovation because even if an individual had the knowledge or ability to adopt an

innovation, an unfavorable attitude towards may result in rejection. Nurses’ attitudes about

genetics and genomics may affect their adoption of this content into their clinical practice.

GGNPS indicators of attitudes/receptivity included the perceived importance of nurses becoming

more educated about genetics of common diseases and perceived advantages and disadvantages

of incorporating genetics and genomics into nursing practice. These were pertinent indicators of

nurses' attitudes because they evaluated the emphasis and priority nurses allocate to

genetics/genomics. The indicators also evaluated both perceived positive and negative aspects of

adopting genetics/genomics into nursing practice. Rogers (2003) stated that attitude is an

important factor because people will adapt a new innovation if they perceive it to have benefits

over currently existing methods. The GGNPS attempts to evaluate if nurses perceive genetics

and genomics as an important part of their clinical role and as something that facilitates their

nursing practice.

Attitudes/receptivity is directly affected by competency/knowledge.

Competency/knowledge were defined as both awareness of the existence of an innovation and

understanding how it functions (Rogers, 2003). Both components are required for adoption of an

innovation. The knowledge an individual acquires about an innovation not only provides insight

into how the innovation functions, but also affects the attitudes/receptivity the individual may

form about the innovation. An individual may have ample knowledge about an innovation, but if

this competency/knowledge leads to negative attitudes/receptivity regarding the innovation, it

may decrease the likelihood of adoption. Therefore, according to Rogers’ DOI model,

competency/knowledge has a direct effect on attitudes/receptivity and an indirect effect on

decision/adoption. GGNPS indicators of competency/knowledge included questions about

collecting and evaluating a family history, the genetic risks associated with common diseases

(such as cancer, diabetes, and coronary artery disease), genetic make-up (similarities in DNA

sequences), and if diseases are caused by a single gene variant. These were pertinent indicators

of competency/knowledge because they considered gene-environment interaction as well as

interactions between genetic variants. In addition, these indicators included knowledge about the

function of genetic variants and how they affected disease processes.

Although competency/knowledge is the first step of the DOI model,

competency/knowledge is affected by additional factors, such as social systems and confidence.

Social systems were evaluated as an independent variable that may support or hinder adoption of

genetics and genomics into nursing practice. This study evaluated if social systems had a direct

effect on competency/knowledge. It also evaluated if social systems had an indirect effect on

attitudes/receptivity and decision/adoption. Social systems were the environment where an

innovation was introduced, such as a clinical site where nurses were employed (Calzone et al.,

2012). Social systems effect adoption of innovations because they are often the sites where

shared decision making and problem solving occurred (Calzone et al., 2012). Therefore, they

may influence acquisition of knowledge; they may also have an indirect effect on

attitudes/receptivity and decision/adoption. Rogers (2003) reports that the structure of a social

system may be a barrier or facilitator to the spread of innovations. He adds that it may be

difficult to evaluate the role of the social system independently from the characteristics of the

individuals within that system (Rogers, 2003). In the GGNPS, social systems were defined as

supervisory support for nurses using genetic/genomics in their clinical practice, as well as

institutional financial support for continuing education in genetics/genomics. These were

pertinent indicators of social systems because they evaluated both support from

management/administration and financial support for adoption of genetics and genomics into

nursing practice.

In addition, this study evaluated if confidence had a direct effect on attitudes/receptivity.

It also evaluated if confidence had an indirect effect on decision/adoption (through

attitudes/receptivity). Although Rogers does not specifically incorporate confidence into the DOI

model, he does consider how personality variables affect adoption of innovations. The DOI

model indicates a direct effect of personality variables on knowledge, and a corresponding

indirect effect on attitudes/receptivity and decision/adoption. Rogers stated that his analysis of

the effect of personality variables on innovativeness is limited, partly because of the difficulty of

“measuring personality dimensions in diffusion surveys” (Rogers, 2003, p. 289). GGNPS

indicators of confidence were the level of certainty in discussing genetics with patients,

considering which family history information is pertinent in evaluating genetic susceptibility to

disease, obtaining current/reliable information, and facilitating referrals for genetic

testing/counseling. These were pertinent indicators of registered nurses' confidence because they

evaluated confidence in the multi-faceted role of the nurse. Registered nurses’ roles and

professional responsibilities include patient education, collection and interpretation of

information (such as a family history), understanding the benefits/limitations of genetic testing,

facilitating referrals to genetic specialists, and working collaboratively within a professional

team. These roles and responsibilities are within the scope of practice of all nurses, regardless of

academic preparation, specialty, or clinical roles (Jenkins & Calzone, 2007). In summary, the

DOI model conceptualizes how innovations, such as genetics and genomics, pass through social

systems and may or may not be adopted into nursing practice. The DOI model supported the

analysis of how the GGNPS evaluates nurses’ competency/knowledge, attitudes/receptivity, the

effects of confidence and social systems, and use genetics and genomics in clinical practice.

Hypotheses/Research Questions

The hypotheses are derived from the relations between the variables of the proposed

model (see Figure 1).

1. Based on baseline data collected at the initiation of the MINC project, the Confirmatory

Factor Analysis results will show that:

a. GGNPS items Part 2, Questions 2.1, 2.2, 2.4; Part 4, Questions 1, 2, 3; Part 5,

Questions 1, 2; Part 6, Questions 1, 2 comprise the underlying latent variable of

COMPETENCY/KNOWLEDGE.

b. GGNPS items Part 1, Questions, 1, 2, 3; Part 2, Questions 2, 3 comprise the

underlying latent variable of ATTITUDES/RECEPTIVITY.

c. GGNPS items Part 3, Questions 2, 3, 4 comprise the underlying latent variable of

DECISION/ADOPTION.

d. GGNPS item Part 2, Question 1 comprises the underlying latent variable of

CONFIDENCE.

e. GGNPS item Part 7, Questions 1.3, 1.4, 1.5, 1.6, and 1.7 comprise the underlying

latent variable of SOCIAL SYSTEM.

2. Based on baseline data collected at the initiation of the MINC project, the above named

latent variables: COMPETENCY/KNOWLEDGE, ATTITUDES/RECEPTIVITY,

DECISION/ADOPTION, CONFIDENCE, and SOCIAL SYSTEM each statistically

significantly influence integration of genetics and genomics into nursing practice in the

hypothesized model based on Rogers’ DOI:

a. COMPETENCY/KNOWLEDGE of genetics/genomics has a statistically significant

direct effect on one’s ATTITUDES/RECEPTIVITY and CONFIDENCE.

COMPETENCY/KNOWLEDGE has a statistically significant indirect effect on one’s

DECISION/ADOPTION of genetics/genomics.

b. ATTITUDES/RECEPTIVITY has a statistically significant direct effect on one’s

DECISION/ADOPTION of genetics/genomics.

c. SOCIAL SYSTEMS have a statistically significant direct effect on

COMPETENCY/KNOWLEDGE, ATTITUDES/RECEPTIVITY, and

DECISION/ADOPTION and a statistically significantly indirect effect on one’s

ATTITUDES/RECEPTIVITY and DECISION/ADOPTION of genetics/genomics.

d. CONFIDENCE has a statistically significant direct effect on

ATTITUDES/RECEPTIVITY, and a statistically significant indirect effect on

DECISION/ADOPTION of genetics/genomics.

3.Does the GGNPS meet the threshold for face validity (as defined by ease of understanding

and ability to apply the instrument to nursing practice by RNs)?

4.Does the GGNPS meet the threshold for content validity based on evaluation by content

experts and analysis of a content validity index (CVI)?

Need for the Study

Genetics and genomics have the potential to change how health care providers screen for,

diagnose, and treat diseases; as well as how we intervene to reduce disease risk. Identification of

genetic variants can be used to determine genetic risk factors and conduct targeted screening; it

can also be used to identify effective pharmacological agents and therapeutic doses.

Understanding and integrating genetics/genomics into clinical practice will allow nurses to

provide evidence-based care to improve patient outcomes; it will also promote evidence-based

use of genetic/genomic information and technology. The GGNPS can be used in educational and

clinical settings to evaluate nurses' use of competencies in genetics and genomics. It is also used

to evaluate registered nurses’ attitudes, receptivity, knowledge, and confidence in genetics and

genomics. Information gathered from the GGNPS can be used to identify knowledge gaps,

create targeted educational initiatives, and/or identify barriers to incorporating genetics and

genomics into clinical practice. In addition, evaluation of nurses’ adoption of genetics and

genomics creates accountability for this content in nursing practice. Genetics and genomics play

an important role in the leading causes of mortality and morbidity and must be a part of

competent, evidence-based nursing practice aiming to improve patient outcomes. The GGNPS

has been evaluated for test/retest reliability; however, validity has yet to be established. As

clinical knowledge advances, new competencies must be added to nursing practice to promote

safe and accountable patient care (Calzone et al., 2014). Validity testing of the GGNPS can

increase its utility as part of a strategic pathway to achieve genetic/genomic competency among

RNs.

Chapter II

Review of the Literature

This chapter discusses how the variables competency/knowledge, confidence,

attitudes/receptivity, and the social system affect decision/adoption of genetics and genomics into

nursing practice. The chapter also explores relations among the variables in the proposed path

model. Specifically, it has been proposed that the adoption of genetics and genomics into one’s

practice is directly dependent upon one’s attitude towards this innovation and the social system’s

support for one’s decision. In turn, one’s attitude is directly influenced by one’s

competency/knowledge, confidence and the social system's support. Finally, one’s confidence is

directly influenced by one’s knowledge and knowledge is directly influenced by the social

system. Due to the pervasive nature of social systems (see model below) its effects are discussed

following each relevant variable.

Decision/Adoption

Figure 2

Variables affecting Decision/Adoption in the adapted model.

Adoption was defined as use of the innovation (Rogers, 2003). For the purpose of this

study adoption was defined as considering and making referrals to genetics specialists, collecting

a personal/family history, and obtaining physical assessment data (Jenkins & Calzone, 2007;

Jenkins, Woolford, Stevens, Kahn, & McBride, 2010). Thus, the GGNPS measured adoption of

Social Systems

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

genetics/genomics into one’s nursing practice, among nurses who actively see patients. Adoption

was measured by self-report of collection and use of family history information within the past

three months. The three-month time frame demonstrates current and consistent use. Information

from the family history was used to provide recommendations to patients and/or to facilitate

clinical decisions, as well as providing the basis for referrals to genetic specialists (Calzone et al.,

2012).

Family history evaluation is a powerful predictor of an individual’s risk of disease

because relatives share genetic makeup and environmental interactions (David et al., 2015). In

fact, in oncology, aside from an early onset of cancer, family history is the most important

indicator of hereditary cancer risk (Weitzel, Blazer, MacDonald, Culver, & Offit, 2011). For

example, a meta-analysis reported that even one family member with pancreatic ductal

adenocarcinoma may increase the patient's risk by 80% (RR = 1.80; 95% CI: 1.48-2.12; n =

6,568) (Permuth-Wey & Egan, 2009).

Similarly, Do and colleagues (2012) used quantitative modeling to show that family

history evaluation is particularly effective for highly prevalent, heritable conditions such as atrial

fibrillation and coronary artery disease. Quantitative modeling included standard liability

threshold models and models of variance in heritability explained by family history and single

nucleotide polymorphism (SNPs). SNPs are differences in DNA sequences based on one

nucleotide (A, T, C, or G); substantial changes in clinical outcomes can result from variation in

such small segments of DNA. The models illustrated that, in common heritable conditions,

family history evaluation may explain approximately 20-30% of disease heritability. This is as

effective as currently available molecular SNP models. For example, coronary artery disease has

more heritability explained by a complete family history (f = 26.3%) than by SNPs (f = 6.9%)

with f as the proportion of heritability explained by family history. In addition, lifetime

morbidity risk is K = 0.402, and heritability of liability is h²L = 0.49 (where h²L indicates the

proportion of variance explained by the substitution of one allele for another allele of the same

trait) (Do et al., 2012). Family history assessment is, therefore, an important tool in clinical

nursing practice to identify increased genetic/genomic risk for a number of diseases.

Furthermore, family history collection is not dependent on availability of technology or

insurance status, and it is possible to obtain in most clinical settings.

All nurses should know when and how to refer clients to a genetic specialist. Referrals

may be considered based on family history results, to provide additional genetic/genomic

information, genetic testing, interpretation of test results, and/or to discuss available services

(American Association of Colleges of Nursing [AACN], 2008). A referral may also be

considered if one or more family members are diagnosed with a condition that may have a

genetic component or for reproductive genetic testing, especially for individuals in certain ethnic

groups (Hampel, Bennett, Buchanan, Pearlman, & Wiesner, 2015). Genetic specialists can also

discuss the benefits or limitations of genetic testing, the results of genetic testing, and possible

genetic causes of diseases.

Shields and colleagues (2010) illustrated the importance of genetic referrals. They found

that maturity-onset diabetes of the young (MODY) is often misdiagnosed as diabetes mellitus

type 1 or 2 because MODY diagnosis can only be confirmed via molecular genetic testing. As a

result, referrals to genetics specialists are vital for correct diagnosis. They reported substantial

positive correlation between prevalence of MODY and rates of referrals for genetic testing.

Clinicians from England referred 1,712 patients for genetic testing and the mutation detection

rate was 27.4% in this group. In contrast, clinicians from Northern Ireland referred only 30

patients and the mutation detection rate was 23.3% in this group (Shields et al., 2010).

Therefore, the rate of detecting mutations was similar in both countries, but increased referrals

for testing increased the number of people who were correctly diagnosed with MODY.

Open, patient-centered communication is another key component of providing quality

nursing care (Lamiani & Furey, 2009). Adoption of genetics and genomics into nursing practice

includes discussing information with patients/families and providing patient education (Calzone

et al., 2012). Registered nurses assess patients’ understanding, perceptions, and beliefs regarding

genetic and genomic information. Nurses can also identify patients who will benefit from

additional information, provide patients with credible, current, and accurate sources of

information, and educate patients about interpretation and use of this information (Jenkins &

Calzone, 2007). Patient education and counseling are vital components of genetic testing

because they are means to inform patients of how a test is conducted, its risks and benefits,

sensitivity/specificity, how results are interpreted and their clinical significance, and whether

follow-up is required.

However, despite the benefits of patient counseling and education, they are performed at

varying frequencies. For example, a study of 39 clinics providing assistive reproductive

technology revealed significant variation of genetic counseling and education practices for

preimplantation genetic diagnosis (PGD) (Girardet et al., 2016; McGowan, Burant, Moran, &

Farrell, 2009). This testing can identify conditions such as Down Syndrome, Huntington’s

Disease, Sickle Cell Disease, and many others. In instances where testing was recommended,

patient counseling was not done consistently. Genetic counseling was required for all such

patients at 56% (n = 22/39) of the clinics, 15% (n = 6/39) of clinics required counseling most of

the time, and 26% (n = 10/39) some of the time (McGowan et al., 2009). Counseling was

provided by certified genetic counselors, physicians, and nurses. The study stressed the

importance of pretest counseling because identification of genetic variants associated with

diseases may lead to significant anxiety among patients. This study also highlights the important

role of nurses in providing genetic counseling because nurses were the most likely to obtain

consent at 17% (n = 6/39) of the clinics for in vitro fertilization procedures, followed by the

patient’s physician at 14% (n = 5/39), or the IVF-specific educator at one (n = 1/39) clinic

(McGowan et al., 2009). Nurses provided genetic counseling and procedure consent for many

patients, supporting patient education, and discussion of genetic/genomic interventions as part of

adoption of genetics/genomics into nursing practice.

Health care providers also play an important role in patient and family education. In

newborn screening for cystic fibrosis and sickle cell disease, nurses and physicians educate

parents about the implications of having a child affected with a condition as well as being a

carrier. Furthermore, health care providers are also a source of education and support when

parents eventually decide to inform their children of their carrier status. A national,

crosssectional, qualitative study in England used semi-structured interviews (n = 67) to explore

experiences of parents receiving positive carrier results for their children following newborn

screening (Ulph, Cullinan, Qureshi, & Kai, 2014). The study also provided insight into parent’s

views of how to inform their children of carrier status. Results indicated that parents wanted to

inform their children to empower them. Many parents focused on reproductive implications and

considered informing their children based on life events such as becoming sexually active.

However, it was difficult for parents to identify a specific age to inform their children. Parents

who were unsure about how to inform their children thought health care professionals would

provide the information. Parents who wanted to inform their children themselves emphasized the

complexity of issues and hoped health care professionals would provide additional education and

support. The study highlighted the important role that healthcare providers, including nurses,

play in ongoing patient and family education regarding the implications of genetic testing.

However, despite the known importance of consistent use of genetics/genomics in nursing

practice, rates of adoption are consistently low. A study using the GGNPS included a

convenience sample of 619 registered nurses, 359 of whom reported actively seeing patients,

self-selected to participate via recruitment on the American Nurses Association (ANA) website;

only 6% (n = 22/359) of participants reported always, and 12% (n = 45/359) often, taking a

family history (Calzone et al., 2013). A follow-up study including 7,798 registered nurses from

17 states and all regions of the U.S. reported similar results. Only 4.1% (n = 204/4,979) of

participants reported collecting a complete family history in the past 3 months; meanwhile,

92.8% (4,563/4,913) reported they rarely or never facilitated referrals to genetics services.

Reported response rates (4,979 and 4,913 respectively) varied due to the number of participants

answering each question (Calzone et al., 2014). Although both of those studies use the GGNPS,

which has not yet been tested for validity, the findings consistently demonstrate low adoption of

genetics/genomics among nurses.

Social systems and decision/adoption. Social systems are proposed to directly affect the

adoption of genetics/genomics into nursing practice (Calzone et al., 2012). Norms of the social

system establish patterns of behavior among its members (Rogers, 2003). According to Rogers

(2003), social systems are interrelated units collaborating in a joint-problem solving effort to

accomplish a common goal. Unit members may be individuals, groups or organizations.

Moreover, social systems serve as a guide or standard for expected behaviors.

Do and colleagues (2012) study concerning the importance of family history taking

highlighted the usefulness of the online, interactive family assessment tools, such as “My Family

Health Portrait,” developed by the Surgeon General and National Human Genome Research

Institute (NHGRI). It is notable that government initiatives often create incentives for health

care institutions to focus on specific topics in clinical care. Do and colleagues point out that this

online tool may guide health care institutions to create a social climate promoting collection of

family history and the interactive format promotes ease of use. The authors also discuss that

even small improvements diagnosing conditions with the use of family history may have

farreaching benefits when applied to a large population. They urged public health initiatives that

allocate resources for using family history disease prevention and screening. Highlighting the

importance of family history evaluation is key to create social system support among health care

organizations.

However, a further challenge for the collection and use of family history information is

electronic documentation; EHRs often do not have a designated location to document and refer

back to family history information (Kho et al., 2011). As a result, nurses are unsure of where to

initially document the information or how to find family history information that is located in an

electronic health record.

Girardet and colleagues (2016) noted that no standardized protocols were available

regarding the use of PGD in reproductive medicine. They suggest that this may be a social factor

adversely affecting standardization of informed consent procedures among clinics because

centers are developing their own clinical procedures. As use of PGD grows, health care

providers are more in need of guidelines from professional organizations. Guidelines could help

create a standard system providing informed consent for patients.

Finally, Shields and colleagues (2010) discussed a number of social system factors that

may prevent correct diagnosis of MODY including the cost of genetic testing. Some health care

institutions may consider genetic testing cost-prohibitive; this may also vary with how

institutions are reimbursed. Additional social system factors include research foci. Areas with a

stronger research interest in MODY have higher rates of diagnosis and referral. Social systems

in those health care organizations may provide additional support because research interests align

with clinical practice. In summary, social systems can directly affect nurses’ adoption of

genetics/genomics through encouraging referrals, family history taking and counseling.

Competency/knowledge is the next variable presented because knowledge about an innovation is

closely tied to its adoption.

Competency/Knowledge

Figure 3

Variables affecting Competency/Knowledge in the adapted model.

Rogers (2003) defined knowledge as occurring “when an individual (or other

decisionmaking unit) is exposed to an innovation’s existence and gains some understanding of

how it functions” (p. 169). Knowledge may be passive or active; individuals may incidentally

discover some information about an innovation and then actively seek information. In the DOI

model shown above, knowledge directly affects attitudes/receptivity; gaining knowledge about

an innovation affects the formation of a favorable or unfavorable attitude towards it. This occurs

because knowledge about an innovation includes information about its existence and information

about how it would be used in an individual’s social system. Deciding to adopt or reject an

innovation is both an information seeking and information processing activity. Thus establishing

the indirect relation between knowledge and decision/adoption, through attitudes/receptivity.

Social Systems

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

Individuals gain knowledge about an innovation, formulate an attitude, and decide to adopt or

reject it. Genetic/Genomic knowledge acquisition ideally begins during basic nursing education.

However, nursing schools have been slow to integrate genetics/genomics into their curricula

(Jenkins & Calzone, 2014; Read & Ward, 2016).

The Health Resources and Services Administration (HRSA) (2000) stressed the

importance of integrating genetics content into nursing curricula to prepare nurses to apply this

knowledge to clinical practice. Despite this, in the five years that followed, only 30% of nursing

schools included such content (Edwards, Quannetta, Maradiegue, Macri, & Sitzer, 2006; Prows,

Calzone, & Jenkins, 2006). A number of factors contributed to this slow adoption into nursing

curricula and corresponding slow adoption into nursing practice: many nurses considered

genetics/genomics a subspecialty and not needed for routine practice; the long list of

competencies at the time was daunting, making it unclear what nurses needed to know;

insufficient numbers of faculty were prepared to teach the content; and state boards did not

consider this knowledge a requirement for licensure or relicensure, therefore there was little

impetus to change curricula (Jenkins & Calzone, 2007, 2014; Read & Ward, 2016).

Currently, the “Essentials of Genetic and Genomic Nursing: Competencies, Curricula

Guidelines, and Outcome Indicators” (2009) are a clear and concise source of competencies

applicable to nursing practice, education, and research. They were used by the American

Association of Colleges of Nursing (AACN) as the foundation for integrating genetics/genomics

into the revised Essentials of Baccalaureate Education for Professional Nursing Practice (2008)

and The Essentials of Master’s Education in Nursing (2011). The AACN Essentials provide the

basis for the Commission on Collegiate Nursing Education (CCNE) accreditation standards for

baccalaureate and masters programs (Jenkins & Calzone, 2014). Despite these developments,

nursing faculty are still struggling to integrate genetics/genomics into nursing curricula.

This is evidenced by a small study (Jenkins & Calzone, 2014), evaluating preparation of

nursing faculty to champion integration of genetic/genomics into their curricula. The study

included a yearlong educational initiative and formation of support networks for nursing faculty

champions to act as change agents at their schools. Prior to beginning the educational initiative,

91% (n =18/20) of the faculty, from all regions of the US except the Northwest, reported a lack

of genetics/genomics knowledge. This is despite the mean of 13.8 years faculty experience

(range of 2-38 years), and the fact that 65% (n = 13/20) of participants had doctoral degrees

(Jenkins & Calzone, 2014).

A larger study, conducted by Read and Ward (2016) included 495 nursing faculty

members from across the U.S also reported a lack genetics/genomics knowledge among nursing

faculty. The faculty completed the Genomic Nursing Concept Inventory (GNCI), a tool

evaluated for reliability and content validity. The GNCI evaluates knowledge related to genome

basics, mutations, inheritance, and genomic health. The mean score was 14.93 (SD = 5.31) or

48% correct. The authors reported that the results were similar to those of nursing students,

mean score of 47% correct. Factors related to a higher score among faculty included doctoral

level education, completion of a genetics course, previously teaching a genetics course, or

teaching genetics content in a related course.

Furthermore, Jenkins and Calzone (2012) reported that, despite agreeing with the

importance of incorporating genetics and genomics into the nursing curriculum, a sample of 167

nurses at AACN and Sigma Theta Tau International meetings reported a limited capacity to

incorporate genetics and genomics into their own curricula. Of the participants, 82% had a

masters or doctoral degree, and 81% were aware that the AACN incorporated genetics/genomics

into the Baccalaureate Essentials. However, 46% rated their genetic/genomic knowledge as

‘low’ and 25% as ‘very low’ (Jenkins & Calzone, 2012). The limited preparedness of faculty

contributes to decreased opportunities for nursing students to gain knowledge about the topic.

Moreover, nursing education does not stop at graduation. Genetics/genomics is a complex and

rapidly evolving science. Nurses need continuing education to maintain current clinical practice

and continuing education is even more important for nurses who have not yet been introduced to

genetics/genomics content. Up to date educational materials are available through the

International Society of Nurses in Genetics (ISONG), the National Human Genome Research

Institute (NHGRI), and the Genetics/Genomics Competency Center (G2C2) (a peerreviewed

resource repository for health care professionals), among others. Peer reviewed publications and

continuing education credits are additional sources of such information.

Despite this, a systematic review by Skirton and colleagues (2012) reported a lack of

genetics/genomics knowledge among practicing nurses. Authors reviewed 269 articles and

included 13, to evaluate nurses’ knowledge, experience, perceived relevance, and confidence

using genetics/genomics in their nursing practice. In none of the 13 reviewed studies did nurses

demonstrate adequate levels of knowledge, as defined by knowledge required to fulfill core

nursing competencies in genetics/genomics. However, one limitation of this review is that

studies varied in how they evaluated nursing knowledge. Thus, though there is a seeming lack of

genetics/genomics knowledge among nurses, it is possible that this may in part be due to the lack

of a valid and reliable instrument to accurately or consistently assess nurses’ knowledge. The

GGNPS can be used to address this research gap by evaluating nurses’ knowledge of

genetics/genomics and promoting knowledge acquisition by identifying areas of knowledge

deficit that can be used to guide educational initiatives.

The GGNPS includes items assessing participants’ knowledge about how to collect a

family history and the role of the nurse in counseling patients about genetic risks. Knowledge

items also assess participants’ perceptions of the relevancy of genetic risk to common diseases

such as breast cancer, coronary heart disease, diabetes, colon cancer, and ovarian cancer. Items

also assess use of family history information and genetic testing to support clinical

decisionmaking. Additional items evaluate nurses’ knowledge of DNA structure, the role of

single gene variants in causing common diseases, and awareness of Essential Nursing

Competencies and Curricula Guidelines for Genetics and Genomics (Calzone & Jenkins, 2011).

It is imperative that there is an awareness of these competencies because they define what all

registered nurses need to know about genetics/genomics, regardless of academic background,

practice setting, role, or specialty (Consensus Panel on Genetic/Genomic Nursing Competencies,

2009).

An additional component of the GGNPS is evaluation of both self-reported and objective

knowledge. Self-reported knowledge is what nurses say they know about the topic. Objective

knowledge is measured by assessing actual knowledge about family history collection and

knowledge of genetic risk associated with common diseases. A deficit of objective measures of

knowledge of genetics/genomics is an important limitation in available research regarding

nurses’ knowledge of this content because self-reported knowledge may differ from objective

measures of knowledge.

For example, Skirton and colleagues (2012) reported that 10 out of 13 reviewed studies

evaluated only self-reported knowledge of genetics/genomics and overall nursing knowledge was

poor. Upon close inspection of the individual studies it was found that, Spruill and colleagues

(2009) reported that 56% (n = 43/77) of nurses rated their genetics knowledge as fair or poor;

while 43% (n = 33/77) reported their knowledge as good. In the study conducted by Tomatir and

colleagues (2006), only 4.2% (n = 2/47) of nurses self-reported that they were knowledgeable

about topics such as autosomal recessive disorders and human genetic diseases, and had

insufficient knowledge about ethical issues in genetics or genetic counseling.

Interestingly, although a very limited number of studies tested nurses’ objective

knowledge of genetics/genomics (as opposed to self-reported knowledge), objective measures of

knowledge often yielded higher scores than self-reported knowledge. In one study, 81.4% (n =

777/955) of nurses chose the correct definition of a mutation and 81.5% (n = 782/960) correctly

answered that men can inherit genetic mutations related to breast cancer (Bottorff et al., 2005).

These authors concluded that self-reported knowledge of genetics/genomics significantly differed

from objective knowledge about the topic. In addition, there were few studies evaluating both

self-reported and objective measures of nurses’ knowledge of genetics/genomics. Thus, the

GGNPS has the capacity to contribute much needed data regarding both self-reported and

objective measures of knowledge of genetics/genomics among nurses.

Social systems and competency/knowledge. Social systems may affect knowledge

acquisition of genetics/genomics among nurses via the emphasis nursing curricula and health

care organizations place on such knowledge. Beginning in 2010, the CCNE accreditation

standards for baccalaureate nursing schools mandated genetics/genomics content based on the

AACN revision of the revised Essentials of Baccalaureate Education for Professional Nursing

Practice (2008). In 2011, the ACCN published the revised Essentials of Mater’s Education in

Nursing, which also included genetics/genomics content. In addition, the AACN noted that

genomics is one of the sciences creating a foundation for nursing practice in the Essentials of

Doctoral Education for Advanced Nursing Practice (2006). As the organizations that set the

standards for nursing education and accrediting bodies continue to highlight the role of

genetics/genomics, this will create greater impetus for curricular change among all levels of

nursing education (Jenkins & Calzone, 2014). However, it is notable that associate degree and

practical/vocational nursing programs are not mandated to include genetics/genomics content,

based on accreditation standards from the NLN-CNEA and NCSBN (“Commission for nursing

education accreditation: Accreditation standards for nursing education programs,” 2016).

Furthermore, NCSBN (2012) is considering a new model of national accreditation for nursing

programs beginning in 2020, therefore additional follow up is required to evaluate if

genetics/genomics content will be included in national accreditation standards.

Jenkins and Calzone (2014) highlighted the role of social systems on knowledge in the

aforementioned nursing faculty champion initiative study. Adoption of the content required, not

only faculty expertise but, support from university administrators and peer networks. For

example, university deans and program directors championed faculty applicants by writing letters

of support and a peer support network was an additional source of social system support.

Faculty champions could discuss questions with nurses who were experts in genetics/genomics.

Support and education from initiatives such as the faculty champion program can help faculty

reach the above-mentioned educational goals outlined by the ACCN and CCNE.

In relation to nursing practice, Kirk and colleagues (2008) identified social systems as a

potential facilitator or barrier to gaining knowledge about genetics/genomics. An example of

social systems facilitating knowledge acquisition is institutions that support continuing education

for nurses. However, in a previously mentioned study Calzone and colleagues (2014) note that,

following licensure, 86.8% (n = 4,594/5,294) of participants reported they had not completed any

courses focusing on genetics/genomics. Allowing time during work hours to complete

genetics/genomics-based courses or educating staff about available courses are potential

facilitators or barriers to knowledge acquisition of this content among nurses.

Competency/knowledge and attitudes/receptivity. Competency/knowledge is

hypothesized to directly affect attitudes/receptivity because knowledge about an innovation

affects the formation of favorable or unfavorable attitudes towards it (Rogers, 2003). Carroll and

colleagues (2009) reported that, an educational intervention positively affected the attitudes of

participants in a small multidisciplinary workshop that included nurses, nurse practitioners,

physicians, and other health care providers. Following the interactive workshop and educational

modules, there was a statistically significant increase in the number of participants agreeing that

genetic testing was beneficial in the management of adult-onset diseases (p= .031), using the

McNemar test for matched categorical variables (Carroll et al., 2009). While the sample size was

small, managers and administrators among those organizations cited the participants as leaders in

their departments. Thus, it is likely that the impact on these leaders’ attitudes towards

genetics/genomics’ usefulness will carry over to their institutions and subordinates.

Similarly, in a previously mentioned study, Calzone and colleagues (2014) empirically

demonstrated the relation between Competency/knowledge and attitudes/receptivity. Nurses (n =

7,798) who attended a genetics course following licensure were more likely to report that

education about genetics/genomics was very important (p <.001, n = 7,798) and strongly agreed

that nurses have a key role in patient counseling regarding genetic risks (Calzone et al., 2014).

Competency/knowledge and decision/adoption. The Calzone and colleagues (2014)

study also reported that educational level (associate to doctorate) affected nurses’ collection of

family history information, demonstrating the relation between competency/knowledge and

decision/adoption, through attitudes/receptivity. Furthermore, nurses who attended a genetics

course following licensure were more likely to ‘always’ or ‘often’ collect family history

information (p <.001, n = 7,798) (Calzone et al., 2014). This may be because the increased

knowledge resulted in favorable attitude towards genetics/genomics and thus increased the

likelihood of their adopting this content into their clinical practice.

In a similar vein but in a non-genetics/genomics context, Cook and colleagues (2013)

studied the effect of an educational intervention on the knowledge, confidence, attitudes, and

clinical practice of nurses caring for patients with mild traumatic brain injury (mTBI). New

clinical recommendations promote ‘brain rest’ and vigilant symptom assessment to promote

recovery. ‘Brain rest’ includes avoiding physical exertion, television, and some electronic

devices. Twenty-eight nurses participated in a quasi-experimental one group pre-test, post-test

study. Knowledge of post-concussion symptoms, symptom assessment, and ‘brain rest’

guidelines were evaluated with an instrument deemed to have content and face validity. An

educational intervention was developed based on the CDC guidelines for head injury.

Following the educational intervention, trauma nurses reported being better able to

identify interventions required to decrease symptoms following an mTBI (χ² = 6.125, p = .001)

(knowledge)(Cook et al., 2013). In assessing nurses’ attitudes, 96% (n = 24/25) of the

participants reported the educational intervention increased their understanding of the importance

of symptom assessment and brain rest guidelines (attitudes). However, while between-group

analysis showed that specialty resource unit nurses (who float between the emergency

department and acute in-patient units) increased their use of mTBI symptom assessment

(5086%), only 52% of the participants reported using mTBI symptom assessment in their clinical

practice (adoption) (Cook et al., 2013). While the study reported a mixed effect of

competency/knowledge on decision/adoption, through attitudes/receptivity, a number of variables

were self-reported and not verified in medical records. Competency/knowledge may also affect

confidence because acquisition of information and ability to apply it may increase confidence.

The following section will explore the variable of confidence and the relation between

competency/knowledge and confidence, as well as confidence and attitudes/receptivity.

Confidence

Figure 4

Variables affecting Confidence in the adapted model.

Calzone and colleagues (2012) defined confidence as the “level of certainty that

knowledge about the innovation is accurate” (p. 12). While Rogers (2003) did not directly

address confidence as a variable that affects adoption of innovations, he did discuss the effect of

personality variables. Rogers stated that earlier adopters are better able to cope with uncertainty

and risk. Benner’s (1984) research about confidence in nursing paralleled many characteristics

described by Rogers. Benner (1984) noted that experienced nurses have more confidence in their

abilities and are more capable of managing difficulties/challenges. As nurses come closer to

becoming experts, they increase their critical thinking ability; critical thinkers exhibit a number

of characteristics including confidence, flexibility, and open-mindedness (Benner, Hughes, &

Sutphen, 2008). This parallels Rogers’s (2003) theory that individuals who can cope with

uncertainty and risk are more likely to adopt innovations. Therefore, this study evaluated the

direct effect of competency/knowledge on confidence as well as the direct effect of confidence

on attitudes/receptivity and the indirect effect of confidence on decision/adoption.

The confidence items in the GGNPS include how confident nurses are about knowing

what family history information is required to evaluate a patient's susceptibility to common

diseases. Additional items include nurses' confidence about discussing how family history

information affects screening intervals, deciding which patients would benefit from referrals to

genetic specialists, and facilitating referrals. Items also concern confidence in the ability to

access reliable and current information about genetics/genomics. Lastly, the GGNPS evaluates

Social Systems

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

how confident nurses are about providing patients with information about the risks, benefits, and

limitations of genetic testing for common diseases.

Calzone and colleagues (2012) reported that only 24% of nurses (n =38/159) were

somewhat or very confident in deciding what family history information is needed to assess

susceptibility to common diseases. Furthermore, Metcalfe and colleagues (2009) reported that,

among hospice nurses, only 3% (n = 3/100) of participants were confident enough to teach

clients and only 20% (n = 20/100) were very confident about explaining how a gene causing

Duchenne Muscular Dystrophy is inherited. These findings are further confirmed in other

nursing specialties outside of hospice care. Andrews and colleagues (2014) noted that many

nurses, in a variety of specialties, lack confidence and this may adversely affect adoption of

genetics/genomics.

Competency/knowledge and confidence. If nurses do not feel confident about their

knowledge, they may not form a positive attitude and thus not provide education about

genetics/genomics, collect family history information, or consider referrals to genetic specialists.

This is true in contexts other than genetics/genomics. In the previously mentioned brain rest

study by Cook and colleagues (2013), researchers studied the effect of an educational

intervention on the confidence of nurses caring for patients with mTBI. The educational

intervention increased nurses' knowledge of symptom assessment and brain test guidelines. It

also increased nurses' confidence in interpreting symptom assessment scores (χ² = 18.05, p <

.001) and providing discharge instructions to patients and family members (χ² = 15.06, p < .001)

(Cook et al., 2013). However, though the educational intervention increased both the nurses’

knowledge and their confidence the authors did not evaluate the effect of knowledge on

confidence, a key relation in the present study.

Silva and colleagues (2016) did take that step. In a non-genetics/genomics study,

researchers evaluated the effect of knowledge on confidence. This effect was measured via an

educational program, about suicide risk assessment and management, on participants' confidence

and level of understanding. The study included 2,843 participants consisting of nurses,

physicians, case managers, administrators, and support staff. Applied Suicide Intervention Skills

Training (ASIST) is a two-day (14-hour total) workshop in which learners are taught to identify

individuals at risk of suicide and how to respond in a way that promotes safety and connects the

individual to community resources. Participants who completed the educational training had

statistically significantly higher confidence in their ability to assess clients and conduct

interventions than participants who did not receive the educational training F (1, 1333) = 298.21,

p < .001.

Confidence and decision/adoption. The previously mentioned systematic review of

nurses’ competence in genetics identified confidence as a main factor in adoption of

genetics/genomics into nursing practice (Skirton et al., 2012). Level of confidence was assessed

for such adoption activities as referring patients for genetic services, discussing genetic

testing/results, and completing a family history. For example, in a study about supporting

parents of children undergoing genetic investigation, Barr and McConkey (2006) reported that

nurses and midwives had low confidence in referring patients to genetic services; of the

participants, only 2.5% (n = 5/194) strongly agreed and 10.8% (n = 21/194) agreed that they felt

comfortable making a referral to a genetic counselor; 1.5% (n = 3/194) strongly agreed and 7.2%

(n = 14/194) agreed that they would feel comfortable discussing test results of pediatric patients

who underwent genetic testing.

Bottorff and colleagues (2005) reported similar findings in another reviewed study

assessing the knowledge, roles, and confidence regarding genetics/genomics among nurses in

Canada. Nurses (n = 975) reported consistently low confidence, ranging from approximately 1.5

– 2.5 on a 5 point Likert scale. Confidence was evaluated for activities reflective of adoption

such as gathering genetic information, providing referrals to genetic specialists, and obtaining

informed consent for release of genetic testing information. Both Barr and McConkey (2006)

and Bottorff and colleagues (2005) noted that decreased confidence may affect nurses’ use of

genetics/genomics in their clinical practice. Literature in this section explored the relation

between knowledge and confidence, confidence and attitudes/receptivity, and the resulting

indirect effect on adoption of innovations such as genetics/genomics. The following section will

explore the variable of ‘attitudes/receptivity.’

This is further supported by findings by Andrews and colleagues (2014), as they

evaluated characteristics of nurses who are more likely to adopt genetics/genomics into nursing

practice. Among a sample of 88 oncology and primary care nurses in the United Kingdom,

adopters were consistently more confident about collecting family history information (mw =

364.5, p < .001) than non-adopters. Researchers created a tool using questions from other

instruments and writing additional questions for this study. Questions from other instruments

were selected based on theoretical underpinning and relevance to this current study. New

questions were evaluated by content experts. The instrument was piloted with a sample of 14

oncology and primary care nurses, but no validity or reliability testing was reported.

Attitudes/Receptivity

Figure 5

Variables affecting Attitudes/Receptivity in the adapted model.

According to Rogers (2003), receptivity is the formation of a relatively enduring

“favorable or unfavorable attitude towards an innovation” (p. 169). Calzone and colleagues

(2012) further define attitudes/receptivity as the perception of advantages of genetics/genomics

as compared to previously used techniques. Therefore, attitudes/receptivity may favorably affect

adoption of an innovation if it is perceived as more advantageous than current clinical practice.

The GGNPS evaluates attitudes/receptivity, including the perceived importance of

becoming more educated about the genetics of common diseases. Calzone and colleagues (2013)

found that the majority of nurses felt that it was very important (67.5%; n = 410/607) or

somewhat important (26.7%; n = 162/607) for nurses to be more educated about the genomics of

common disease. In a previously discussed study, Carroll and colleagues (2009) report that,

when members of a small multidisciplinary workshop were asked about their attitudes toward

genetics/genomics, 95% (n = 20/21) reported that they perceived the content as important.

Likewise, Coleman and colleagues (2014) reported that 79% (n = 301/383) of participants felt it

was ‘very important’ for nurses to be more educated about genetics/genomics. This study also

used the GGNPS but included a diverse sample of nurses (Black/African American 33%, Asian

43%, Hispanic/Latino 18%, White 8%, American Indian 1%, Native Hawaiian/Pacific Islander

3%, and Other 12%), while many nursing studies underrepresent minorities. Overall, both of

these studies consistently reported that nurses had a positive attitude concerning the importance

of genetics/genomics to nursing practice.

Social Systems

Comp

etency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

The GGNPS also evaluates nurses’ perceptions of the potential advantages and

disadvantages of using genetics and genomics to support treatment decisions, improve service to

patients, and promote patient adherence to clinical recommendations. The attitudes section of the

GGNPS also evaluates nurses’ perceptions of the clinical advantages of using genetics/genomics.

An example of a major clinical advantage of genetics/genomics is its use in treatment selection to

promote effectiveness and decrease risks of adverse events. Evaluating the relation between

genetics/genomics and medication safety, Kelly and colleagues (2012) published case reports of

two deaths and one event of respiratory depression in children aged 3 to 5 years. Though all

three children received appropriate doses of codeine after tonsillectomy and/or adenoidectomy,

genetic variation resulted in ultra-rapid metabolism of codeine into morphine increasing the risk

for toxicity and death. Patients’ postmortem morphine levels were significantly higher than

therapeutic doses. In response to this information and following additional investigation, the

FDA (2017) issued an updated safety announcement indicating that codeine and tramadol should

not be administered to children under 12 years of age.

Research about the safety of codeine administration demonstrates the clinical advantages

of genetics/genomics to nursing practice and the following research by Metcalfe and colleagues

(2009) illustrates how nurses’ attitudes about such advantages can change. Researchers initially

reported that 43% (n = 43/100) of nurses felt that integrating genetics into hospice care was of

‘little importance’. However, as they progressed through reading the scenarios concerning

applications of genetics in hospice care, two thirds, up to 99%, rated genetics as ‘very important’

or ‘essential’ to palliative care (Metcalfe et al., 2009). Thus, knowledge changed their attitudes

towards advantages, to patients, of using genetics/genomics in their practice.

The GGNPS also evaluates attitudes towards the perceived disadvantages for nurses of

using genetics and genomics, such as a lack of time, reimbursement difficulties, and/or the need

to obtain a new skill set. Inadequate time to complete a high workload is a frequently reported

stressor among nurses. Tourangeau, Thomson, Cummings and Cranley (2013) reported that the

reasons for nurses leaving in-patient jobs included inadequate staffing (68.8%, n = 2718/3950),

unmanageable workload (62.1%, n = 2454/3950), and “inadequate time to do job well” (43.8%, n

= 1730/3950). Correspondingly, nurses report inadequate time as a major barrier to collecting

family history information, further demonstrating that high workload and inadequate time are

perceived disadvantages of genetics/genomics (Sussner, Jandorf, & Valdimarsdottir, 2011).

In a non-genetic/genomic context, but in a similar vein, a qualitative study by Lee (2004)

also highlighted time pressures perceived by nurses. This study used Rogers’ (2003) DOI model

to evaluate nurses’ perceptions of the innovation of computerized care systems. Interviews were

obtained from 12 nurses working in three respiratory intensive care units. Time shortage was

identified as a theme. Additional time required to use the computerized system was a stressor

that negatively affected nurses’ attitudes toward the innovation. They reported the perception

that management valued documentation more than patient outcomes.

Finally, the GGNPS evaluates nurses’ perceived disadvantages for patients. A potential

disadvantage is that receiving genetic testing results may be anxiety-provoking for patients. A

meta-analysis, by Hamilton, Lobel, and Moyer (2009), evaluated anxiety and distress related to

BRCA1/2 mutation testing. Twenty studies, totaling 2,171 participants (95.2% female), were

included. There were statistically significant differences in anxiety reported by carriers,

noncarriers, and individuals with inconclusive results after a short period of time (mixed effects

Q (2) = 22.41, p < .001); but no statistically significant difference in anxiety after a moderate

period of time (mixed effects Q (2) = 2.47, p =.29) (Hamilton et al., 2009). Patients may need

additional support and counseling in the period directly following genetic testing when their

anxiety may be elevated. Nurses can provide information about the tests results, as well as

limitations of currently available testing, and facilitate referrals, if warranted. In addition, nurses’

attitudes/receptivity about genetics/genomics may be affected by patient experiences of anxiety

or distress upon receiving genetic testing results.

An additional potential disadvantage of genetics/genomics for patients is loss of privacy

and fear of discrimination. A qualitative review of literature on the adoption of online therapy

among psychologists identified patient privacy and confidentiality concerns as potential barriers

to adoption of and attitudes toward online therapy by psychologists (Lovejoy, Demireva,

Grayson, & McNamara, 2009). These concerns parallel concerns about discrimination based on

genetic information among patients. Online services present challenges of safe record keeping

and HIPAA compliance. The use of passwords can provide security, but passwords may be

shared or intercepted by others. The researchers stressed reporting to clients that online sessions

have limits of confidentiality and there are associated risks and benefits. In addition, providers

need to obtain information prior to initiating therapy (such as name and address) in case of an

emergency. Concerns about loss of privacy and fear of discrimination by patients were a barrier

to adoption of the innovation by the health care providers. Privacy concerns by patients may

affect nurses’ attitudes towards adoption of genetics/genomics. Educating nurses about the

Genetic Information Nondiscrimination Act (GINA) of 2008 can correct misconceptions about

lack of privacy and address the fear of discrimination.

Attitudes/receptivity and decision/adoption. Nurses’ attitudes towards

genetics/genomics are hypothesized to directly affect its adoption into one’s clinical practice. It

was noted that nurses are time stressed and do not have time to do their jobs properly and

perceive this as a deterrent to aspects of their practice that are not mandated by administration

(Tourangeau et al., 2013). Jenkins and Calzone (2007) reported that evaluation of a family

history by a registered nurse is not reimbursed. Nurses have many time-sensitive roles and

responsibilities required for continued employment in an organization. If medications are not

administered within the required time period, many health care documentation systems require

written explanations about why this occurred. Some health care institutions also monitor call

bell response time (Roszell, Jones, & Lynn, 2009). Unlike medication administration and call

bell response times, employers do not currently monitor collection of family history information.

In addition, nurses often do not know how to use the information after it is obtained or where to

document it in the patient’s chart. This may contribute to Calzone and colleagues (2014)

findings that, although nurses recognized the importance of genetics and genomics in clinical

practice, actual adoption of this innovation is low.

Confidence and attitudes/receptivity. In the present study, it is hypothesized that

confidence has a direct effect on one’s attitude/receptivity. Calzone and colleagues (2013) noted

that only 27.5% of surveyed nurses reported being confident or very confident about choosing

what family history information was needed to assess genetic susceptibility to common diseases.

Given such low confidence, it was expected that the same surveyed nurses would report an

unfavorable attitude towards use of genetics/genomics. Interestingly, 94.2% of surveyed nurses

felt that it was very important or somewhat important for nurses to become more educated about

genomics of common disease, thus demonstrating a positive attitude.

While this study seems to indicate that low confidence may not directly influence nurses’

positive attitudes/receptivity regarding genetics/genomics, nurses’ attitudes towards the

importance of nurses learning more about genetics/genomics are only one piece of the GGNPS

attitudes section (Calzone et al., 2013). The other attitudes items assess nurses’ perceptions of

advantages and disadvantages to both themselves and their patients. Furthermore, as detailed

below, knowledge and social systems also affect nurses’ attitudes. It is thus likely that, one or

both of these factors may have a stronger effect on attitudes than does confidence alone.

Concerning the influence of knowledge Metcalfe and colleagues (2009) evaluated the

confidence and attitudes of hospice nurses regarding the use of genetics in their clinical roles. As

previously noted, initially the attitude of 43% (n = 43/100) of the nurses was that integrating

genetics into hospice care was of ‘little importance’ and only 3% (n = 3/100) were confident

enough to teach clients. However, after reading scenarios about applications of genetics in

hospice care, two thirds, up to 99%, rated genetics as ‘very important’ or ‘essential’ (Metcalfe et

al., 2009). Perhaps it was not just confidence that affected attitudes but also the knowledge

gained through exposure to relevant clinical scenarios.

Social systems and attitudes/receptivity. Regarding social systems’ influence on

nurses’ attitudes/receptivity about genetics/genomics the most frequently reported stressors

among nurses include inadequate time, inadequate resources, an unsupportive organization,

inadequate staffing, and a poor work environment (Tourangeau et al., 2013). The addition of

genetic/genomic responsibilities to nursing practice results in even more time pressure (Sussner

et al., 2011) and can negatively affect their attitude towards these responsibilities.

Genetics/genomics may create additional time pressure because nurses frequently report

inadequate time to complete their current workload and would have additional responsibilities

such as collection of family history information, patient education, and facilitating referrals to

genetic specialists.

Likewise, the support of supervisors or managers likely affects nurses’ attitudes about

genetics/genomics. In the previously discussed study by Calzone and colleagues (2014), only

25.3% (n = 1,342/5,314) of participants reported they believed their senior staff members

considered genetics as an important part of the participants’ role. In addition, 45.5% (n =

2,430/5,343) reported they could not attend a genetics/genomics course during work hours.

This may be due to the fact that, registered nurses and health care organizations are not

reimbursed if nurses collect or evaluate family history data (Jenkins & Calzone, 2007). Nor is

their reimbursement jeopardized by nurses’ failure to collect and evaluate this data. Beginning in

2012, the Center for Medicare and Medicaid Services began using the Hospital Consumer

Assessment of Healthcare Providers and Systems (HCAHPS) patient satisfaction survey to

determine reimbursement of 1% of diagnosis related group (DRG) payments, or $850 million

dollars (Zusman, 2012). This number is likely to increase in the coming years as private insurers

adopt similar strategies. This creates a social pressure among hospital administrators, and

therefore nurses, to prioritize care that is reflected on HCAHPS, such as prompt call bell

response time. By contrast, collecting family history information and facilitating referrals to

genetic specialists are not items evaluated on HCAHPS or reimbursed for registered nurses.

Thus, pressure to complete many required responsibilities, lack of supervisory support, and a

lack of reimbursement create potential social systems-related pressures that can influence nurses’

attitudes towards genetics/genomics use in their own practice. Reimbursement affects not only

nurses’ attitudes towards genetics/genomics, but also the affect social systems may have on

adoption. Health care organizations may prioritize care that increases reimbursement to meet

budgetary needs and increase revenue (Trinkoff et al., 2011). However, it is important for health

care organizations to consider value-based care. This health care model focuses on positive

patient outcomes per dollar spent; to contain costs (by decreasing unnecessary tests and

procedures) and prioritize effective patient care. While hospitals may feel pressured to increase

reimbursement, ultimate responsibilities of health care institutions are to use available resources

(such as genetics/genomics) to improve patient outcomes.

Conclusion

Genetics/genomics is used in all aspects of clinical care; because it is so widely used in

health care, adoption of genetics/genomics is required to provide safe, evidence-based, and

competent nursing care. A valid and reliable instrument evaluating nurses’

competency/knowledge, confidence, attitudes/receptivity, decision/adoption of this content, as

well as the effect of social systems, can be used to identify gaps in knowledge, deficits in

confidence, nurses’ attitudes and perceptions of the of the use genetics/genomics in clinical

practice, potential barriers, and ultimately adoption of the content into nursing practice. This

information can be used to create continuing education initiatives, changes in curricula, clinical

competencies, and/or policy initiatives.

Chapter III

The Method

The purpose of this study was to evaluate the Genetics and Genomics in Nursing Practice

Survey (GGNPS) instrument for construct, content, and face validity. This chapter presents the

study method and is organized into five sections: (a) design, (b) sample, (c) data collection

procedures, (d) instrument, and (e) data analysis.

Design

In this secondary analysis of data, a content validity ratio was calculated from a content

validity index. Eight nursing experts were selected based on current publications and clinical

expertise. Some of the experts were clinicians providing care to patients with genetic conditions,

others researchers with expertise in genetics/genomics, and others experts in ethical, legal, and

social implications (ELSI) of genetics. Content validity evaluation informed this analysis via

feedback from nursing experts about each individual item in the GGNPS, as well as via a

comprehensive CVI score for the overall instrument.

A Confirmatory Factor Analysis (CFA) was conducted to test the fit of a factor model,

based on Rogers’ DOI model, to data from GGNPS surveys administered to a nationwide sample

of Registered Nurses (RNs) (Calzone et al., 2016). A CFA was used because it evaluated the fit

between a theory-proposed model and empirical data (Comrey & Lee, 2013). The latent

variables were created based on mapping the GGNPS items to the DOI model. The latent

variables used in this analysis were identified using conceptual and operational definitions

designated by Calzone and colleagues (2012), who based the variables and construction of the

GGNPS on Rogers’ DOI model (Rogers, 2003).

Construct validity was also analyzed via structural equation models (SEMs). The SEMs

tested model fit and the relations between the factors described in Rogers’ DOI model. The

ultimate goal of the SEMs was to evaluate how well the proposed model predicted adoption of

genetics/genomics into nursing practice.

Following the secondary data analysis, new participants were recruited to evaluate face

validity. Face validity was assessed via feedback on the GGNPS from seven currently practicing

RNs. Participants included nurses with one year or more clinical experience, working in a

variety of practice areas (acute and community settings). Clinical competence as an RN was

determined by current employment and duration of nursing experience (one year or greater). To

prevent sampling bias, nurses were recruited from a variety of clinical backgrounds. Participants

were practicing RNs with an associate, bachelors, master’s, or doctoral degrees. Participants had

a variety of clinical experience, ranging from 2 to 31 years of nursing practice. Face validity

informed the analysis by providing insight into how the instrument is understood and interpreted

by participants.

Sample

Calzone and colleagues (2014) collected GGNPS data from 7,798 RNs from 17 states,

representing all regions of the U.S. At the time of the survey, the nurses were employed at one of

23 Magnet® Recognition Program hospitals. Institutions included acute care centers, three

children’s hospitals, a Veterans Administration, one rural hospital, one psychiatric hospital, and

one cancer center. Nurses participated in a one-year study to evaluate nurses’

competency/knowledge, confidence, attitudes/receptivity, and the effects of social systems on

their adoption of genetics/genomics into their clinical practice. Data were collected at baseline

and following a one-year competency integration initiative. The baseline data was used for this

study. Baseline data was selected because it is most likely to reflect competency/knowledge,

confidence, attitudes/receptivity, effect of social systems, and adoption of genetics/genomics

among RNs within the U.S., as opposed to RNs who participated in an initiative to incorporate

genetics/genomics into their clinical practice. Eligibility criteria for the sample included being a

RN currently employed at the health care center at the time of survey administration. RNs of all

academic backgrounds were eligible to participate.

Data Collection Procedures

Institutional review board approval for this study was sought from the College of Staten

Island (CUNY). Exempt status was sought because data were not collected with identifying

information. Research using pre-existing data is exempt from institutional review board

evaluation if information is recorded in a way that participants cannot be directly or indirectly

identified. After obtaining exemption, the investigator downloaded the dataset onto a

passwordprotected computer; because there is no direct involvement with human subjects and

will be conducted on previously collected and de-identified data, no harm to human subjects is

anticipated.

Calzone and colleagues (2014) originally collected the data following approval from the

Institutional Review Board (IRB) of West Virginia University (WVU). A reliance agreement was

also established between WVU and the Office of Human Subjects Research at the National

Institutes of Health (NIH); some institutions relied on the WVU-IRB approval, others required

additional IRB review by their institutions.

The GGNPS was administered at each health care institution between July and October

2012 as part of a multi-institution project called Method for Introducing a New Competency into

Nursing Practice (MINC) (Calzone et al., 2014). The project included 23 hospitals; baseline data

was collected from 7,798 RNs. Response rates varied from 17% to 63%. The survey was open

for completion for 28 days. Recruitment was conducted via email notification; however,

institutions could utilize additional recruitment strategies such as advertising, supervisor

encouragement, intranet posting, and incentives in accordance with local IRB approval. The

baseline data collected at the initiation of the MINC project was used for the statistical analysis in

this study.

Instrument

Origins. The GGNPS was derived from an instrument used to evaluate components of

the DOI model: attitudes, knowledge, competency, receptivity, confidence, and adoption of

genetics/genomics by family physicians (FP’s), as well as the effect of social systems (Jenkins et

al., 2010). It was developed by a team consisting of FPs, survey designers, a behavioral scientist,

and experts in genetics/genomics. Content validity for the FP instrument was evaluated by

content experts. A confirmatory factor analysis (CFA) was done on data from the pilot study,

individual factor loadings were not reported in the article.

Construct validation was conducted using structural equation modeling (SEM). SEM was

used to evaluate whether the items in the instrument aligned with the DOI domains and evaluate

the paths among the variables. The results of the SEM for the FP instrument included: Tucker

Lewis index (TLI=.92, > .9), root mean square error of approximation (RMSEA=.039;

90% confidence interval, < .08), and standardized root mean square residual (SRMR=.056, < .1)

(Jenkins et al., 2010). Jenkins and colleagues (2010) report that the findings indicate the

instrument is aligned with the domains of the DOI.

While the standardized path weights were statistically significant (p < 0.001), Jenkins and

colleagues (2010) noted two pathways leading to adoption of genomic-related innovations, as

opposed to one pathway as predicted by Rogers’ DOI model. SEM was used to evaluate

pathways that influenced FP’s adoption of genomic-related innovations. In pathway 1, adoption

of genetics/genomics into clinical practice by FP’s was associated with increased comfort with

using genetic information. In pathway 2, adoption was associated with perceived relevance of

family history and genetics of common diseases information to clinical practice.

GGNPS development. The FP instrument was adjusted for nursing practice. To do this,

Calzone and colleagues (2012) consulted with nursing and genomic nursing experts. The items

were evaluated for nursing scope of practice, consistency with genomic competencies for nurses,

and for content validity. The content experts reviewed the proposed instrument and provided

feedback and suggested changes. Though the provided feedback was not published, content

experts agreed with edits made to the FP instrument, revising it for nursing scope of practice.

Edits included changing items about referring clients to genetic/genomic services to “facilitating

referrals” and medication administration as opposed to prescribing medications.

The GGNPS was first pilot tested for usability by a convenience sample of five nurses

and tested again using a larger sample (Calzone et al., 2012). A second pilot test was conducted

on a larger sample of 239 RNs employed at the NIH, response rate was 28%. The GGNPS was

amended following pilot testing. Amendments were based on feedback, follow-up review by

experts, and recommendations from pilot test participants. Amendments included the removal of

questions that expert reviewers considered unclear and addition of two questions about

knowledge of genetics of common diseases from the Genetic Variation Knowledge Assessment

Index (GKAI).

GKAI. The GKAI is an instrument evaluating health care professionals’ knowledge of

genetic variation and race. The GKAI was evaluated for validity via confirmatory and

exploratory factor analysis, expert advisory panels, and focus groups (Bonham, Sellers, &

Woolford, 2014). Two questions pertaining to genetic/genomic knowledge were added to the

GGNPS with permission from the developers and authors.

Reliability testing. Reliability testing was conducted on data from two institutions

participating in the MINC study (Calzone et al., 2016). The MINC study included a baseline

assessment and an assessment at the completion of the one-year educational intervention.

Reliability testing was conducted following the educational intervention; those participants

completed the instrument twice to assess for test-retest reliability. At Time 1, the GGNPS was

available electronically for completion for 28 days. One week later the survey was opened for 14

days for Time 2 (retest). Participants were provided a personal study number to match responses

for both time periods. If responders did not have the same number for Times 1 and 2, had a

missing number, or a duplicate number with another participant, then they were excluded from

the study. A personal study number was used in place of identifying data to promote

confidentiality. A total of 993 RNs completed the GGNPS at Time 1 and 564 at Time 2.

However, after excluding responses that were missing a personal study number, used a duplicate

number, or had all data missing for Times 1 or 2, 232 responses were used for analysis.

During reliability testing, Likert scale items with five or more options performed poorly

and the questions were amended, by shortening the scale, to provide more accurate results in

future testing. Moderate agreement was classified as a Kappa of 0.41-0.60 (Landis & Koch,

1977). Moderate agreement was achieved in 39% (n = 36/95) of instrument items. The mean

agreement across all items in the instrument was a Kappa of 0.407 (range 0.150-1.000) (Calzone

et al., 2016).

‘Select-all-that-apply’ items performed the poorest, mean K=0.308 (range 0.165-0.432)

for questions pertaining to attitude/receptivity and mean K=0.262 (range 0.150-0.394) for

questions pertaining to self-assessment of ability to discuss genetics of common diseases with

clients (Calzone et al., 2016). Among the Likert scale items, questions with 7 response options

performed poorly, mean Kw=0.324 (range 0.216-0.375), while Likert scale items with 4 or 3

response options performed better (Calzone et al., 2016). Of the two items that were revised, one

was in the Competency/knowledge domain and one in Attitudes/Receptivity.

The version of the GGNPS used for this study did not include amendments made

following reliability testing, because reliability testing was completed after baseline data was

collected in the MINC study. The baseline data sample was a robust nation-wide sample well

suited for statistical modeling. The analysis considered how amendments to the GGNPS,

following reliability testing, may have affected the statistical analysis. Future research will

include redistributing the revised instrument and comparing the SEM results.

Content. The GGNPS includes nominal (such as true/false and yes/ no questions) and

ordinal questions. The GGNPS domains include decision/adoption, competency/knowledge,

attitudes/receptivity, confidence, and the effects of social systems. The decision/adoption domain

includes four questions about how often, in the past three months, have participants collected a

complete family history, cared for a patient who initiated a discussion about genetics/genomics,

facilitated a referral for genetic services, and used family history information to facilitate clinical

decisions or patient recommendations. Collection of family history information is used as a

measure of adoption because it is not dependent on technology or financial resources. The

competency/knowledge domain includes 20 questions asking participants to agree/disagree or

mark “don’t know” for statements about how family history is collected, such as which relatives

are included. The competency/knowledge domain also includes questions about genetic risk

associated with breast cancer, colon cancer, coronary heart disease, diabetes, and ovarian cancer.

Additional questions include which information is collected when patients indicate a disorder in

the family (age of diagnosis, relation to the patient, race/ethnicity, age of death from condition,

information from both maternal/paternal lineages), as well as questions about DNA structure,

gene variants associated with diabetes/heart disease, and awareness of the Essential

Competencies and Curricula Guidelines. The final questions ask participants to evaluate their

understanding of genetics of common disease through self-report. Thus, questions in the

competency/knowledge domains evaluate both self-reported knowledge and objectively test

nurses’ knowledge of genetics/genomics, providing both subjective and objective measures. The

attitudes/receptivity domain includes ten subjective questions about the potential advantages and

disadvantages of integrating genetics/genomics into clinical practice, the perceived importance of

becoming more educated about the genetics/genomics of common diseases, and taking a family

history as a component of nursing care. The confidence domain includes eight questions about

nurses’ self-reported confidence regarding what family history information is needed to evaluate

susceptibility to common diseases, discussing how family history affects screening for diseases,

which patients may benefit from referrals for genetic counseling and testing, providing patients

with information regarding risks/benefits of genetic testing, and facilitating referrals for genetic

services. The social system domain includes five questions about whether participants feel that

senior staff members view genetics/genomics as an important part of a nurse’s role as well as

their own role. Additional questions ask if participants would attend genetics/genomics classes

on their own time, could take a class during work hours, and if they intend to learn more about

genetics.

In scoring the GGNPS, items from the confidence, attitudes/receptivity, social system,

and adoption domains were analyzed independently and not combined to form an overall score.

The twelve competency/knowledge items were combined to form a knowledge score, making

knowledge a continuous variable (see Appendix B). Responses to the twelve-

competency/knowledge items were marked as correct (1) or incorrect (0). If participants

responded “don’t know” that was also marked as incorrect (0), the number of correct answers

was added, generating a score from 0 to 12. Calculation of a total competency/knowledge score

was restricted to surveys where all 12 questions were answered.

Data Analysis

Data was downloaded into Microsoft Excel 2010 spread sheets. It was analyzed using

Mplus version 7.3 and R version 3.1.2. Data was analyzed in three steps. GGNPS items are

nominal, ordinal, and continuous variables. As a result, finite mixture models were used to

evaluate the heterogeneity and underlying distribution among the latent variables. The data

analysis included attempting to measure the CFA of a model comprised of Rogers’ DOI factors:

competency/knowledge, confidence, attitudes/receptivity, social systems, and decision/adoption.

SEMs were conducted to evaluate the overall model fit and the relation between the

factors; relations were evaluated based on Rogers’ (2003) DOI model. Chi-square fit tests were

used to determine the fit of the proposed path models to the data and to compare the relative fit

of the models with each other (Hoyle, 1995). Consideration was given to the limitation of the

chi-square test with a large sample size; statistical tests such as RMSEA, BIC, and CFI were also

evaluated (Browne & Cudeck, 1993).

Content validity was calculated using a content validity ratio. Eight experts in nursing

and genetics/genomics were sent a content validity index derived from the GGNPS. Experts

were defined as individuals who have current employment, experience, and evidence of

scholarship in both nursing and genetics/genomics. Furthermore, experts with backgrounds in

genetic/genomic research, education, and clinical practice (as opposed to solely bench science)

were selected, because the GGNPS is a tool for nurses to apply genetics/genomics to their

clinical practice. The content validity index was created from each item in the GGNPS with the

definition of each domain and corresponding items provided for the content experts. A relevance

table with four categories (ranging from ‘very relevant’ to ‘not relevant’) was used to score each

item (Lawshe, 1975). The content validity ratio was calculated using Lawshe’s formula:

CVR = [Ne – (N/2)]/ (N/2)

In this formula Ne was the number of panelists who classified the item as ‘very relevant’ or

‘relevant’ and N is the total number of content experts. Revisions, based on the CVR and the

overall CVI score, were proposed.

An exploratory factor analysis (EFA) was conducted because the CFA was unable to

converge on a stable set of parameter estimates. The items that were hypothesized to make up

the variables of competency/knowledge, attitudes/receptivity, confidence, decision/adoption, and

social systems could not be grouped as hypothesized when the CFA analysis was conducted.

Iterative maximum likelihood calculations were unable to fit the proposed variables to the data.

A follow-up EFA analysis indicated that this may be due to items being related to variables

different than those that were originally hypothesized. In addition, some items may have

grouped together due to similar formatting (e.g. confidence items).

In an EFA, factor loadings are derived from relations among the data; the data-driven

EFA compliments the theory-driven CFA, based on work by Calzone and colleagues (2012) and

Rogers’ (2003) DOI theory. The EFA provided additional insight into which items comprised the

variables of competency/knowledge, attitudes/receptivity, confidence, socials systems, and

decision/adoption.

Chapter IV

Results

The purpose of this study was to test the content, face and construct validity of the

Genetics and Genomics Nursing Practice Survey (GGNPS) (Calzone et al., 2012). A content

validity ratio (CVR) was first calculated on data from seven content validity experts. Then face

validity was evaluated via feedback from seven currently practicing RNs. Finally, construct

validity was evaluated in two ways using existing data from GGNPS surveys administered to a

nationwide sample of RNs (Calzone et al., 2016): 1) a Confirmatory Factor Analysis (CFA)

tested the fit of a factor model, based on Rogers’ DOI model, to these data; 2) A series of

structural equation models tested the fit between hypothesized relations described in Rogers’

DOI model and these data. An Exploratory Factor Analysis (EFA) was conducted to gain insight

about why the CFA was unable to converge and to explore relations among the items in the

GGNPS to inform future analyses.

The GGNPS The GGNPS measures nurses’ competency/knowledge, attitudes/receptivity,

confidence, and decision/adoption of genetics and genomics, as well as the effects of social

systems on adoption. The format includes a mix of multiple-choice, dichotomous, and Likert-

scale questions. A knowledge score was calculated by totaling the number of correct responses

to the 12 knowledge items (knowledge scores range from 0 to 12). Cumulative scores were not

calculated for the attitudes/receptivity, confidence, social system, and decision/adoption

variables; this is because some items in those variables were responders’ subjective opinions

(e.g. “Please indicate whether you agree or disagree with the following statements”). The

instrument also includes demographic data.

Data Collection Results and Response Rates

This is a secondary analysis of data collected as baseline data (time 1) as part of Calzone

and colleagues’ (2014) yearlong intervention aimed at improving nurses’ capacity to integrate

genetics and genomics into their clinical practice. RNs from a total of 23 Magnet® designated

hospitals, across 17 states, participated in this study. In those 23 hospitals, RNs with any role (n

= 27,613) were eligible to participate. Response rates of eligible participants varied from 1763%

between hospitals. The GGNPS was administered between July and October 2012. It was open

for completion at each institution for 28 days. Institutions sent an email notification of the

survey availability and reminder emails as well as using additional recruitment strategies such as

advertising, intranet posting, and supervisor encouragement.

All participating hospitals reported similar barriers to completion, such as burden due to

time constraints of practicing nurses; 100% of the institutions were concerned with survey

burden and had conducted institution-wide nursing surveys in the past six months. A total of

7,842 responses were received. Of these, Calzone and colleagues (2014) excluded 12 non-RN

respondents and an additional 32 respondents who did not report hospital affiliation. This

resulted in 7,798 eligible respondents and an overall 28% response rate.

Missing Data

Of these 7,798, 8.71% ( n = 679) respondents left all items blank and were excluded,

leaving 7,119 respondents. Figure 6 contains a frequency distribution of the percentage of the

GGNPS participants completed (range 0-100%). Demographic items provide important

information about participants, such as clinical experience. The sample was first inspected for

respondents who answered most demographic items, but omitted some answers to study

variables. A ratio was calculated of demographic items answered to total items answered.

Individuals who had a z-score of +/-3 or greater were excluded (n = 3). These individuals

answered a majority of demographic items, without providing information pertinent to evaluation

of the study variables. This left 7,116 respondents.

Figure 6

Percent of GGNPS Completed by Participants.

When evaluating the relations between study variables, analyses may be affected if items

are answered mostly for one study variable, but not others. Thus, respondents who answered the

majority of items pertaining to one variable (e.g., social systems), but not to the other variables,

were removed. Ratios between complete data on individual variables versus the entire

instrument were calculated (see Table 1); individuals who had a z-score of +/-3 or greater were

excluded. Notably, excluded individuals answered the confidence and/or decision/adoption items

which were grouped in the beginning of the GGNPS. Thus, these individuals likely answered

items in the beginning, then left an increasing amount of items blank as they proceeded through

4974

2000

4000

6000

Percent Complete

the GGNPS. Seven individuals had a z-score of +/-3 in relation to both confidence and

decision/adoption as compared to the entire instrument.

Table 1 z-Scores +/- 3 for All

Variables.

Ratio

Number of Participants with z-Score +/- 3

Social Systems to total

GGNPS Complete

Competency/Knowledge to

total GGNPS Complete

Confidence to total GGNPS

Complete

127

Attitude/Receptivity to total

GGNPS Complete

Decision/Adoption to total

GGNPS Complete

134

Data from the remaining 6,861 respondents were then examined for missing data patterns.

Prior to evaluating whether data were missing at random (MAR) (Tabachnick, Fidell, &

Osterlind, 2001), normality tests were conducted to determine whether parametric or

nonparametric tests should be utilized to evaluate missing data. The Anderson-Darling and the

Cramer-von Mises (Razali, Wah, 2011) tests demonstrated statistically significant deviations (p <

.05) from a normal distribution. While this indicates that the data may not be normally

distributed, this is an expected result due to the large sample size (see Tables 2 and 3). Q-Q

plots, a graphical representation of normality, suggest that deviations from normality were small

and tended to occur near the tails of the distribution (see Appendices C.1-C.5); however, the

confidence and social system variables demonstrated truncated tails, implying range restriction.

Because the deviations of the data from normality appeared small, parametric tests were used to

evaluate whether data were MAR.

Table 2

Cramer-von Mises Normality Test.

Variable

Cramer-von Mises Normality Test Value

p-Value

Social Systems

18.577

p < .001

Competency/Knowledge

.89781

p < .001

Knowledge Score (0-12)

.06666

p = .3093

Confidence

6.5058

p < .001

Attitude/Receptivity

5.4833

p < .001

Decision/Adoption 2.4749 p < .001

Table 3

Anderson-Darling Normality Test.

Variable

Anderson-Darling Normality Test Value

p Value

Social Systems

118.31

p < .001

Competency/Knowledge

5.1671

p < .001

Knowledge Score (0-12)

.38739

p = .3877

Confidence

46.437

p < .001

Attitude/Receptivity

31.891

p < .001

Decision/Adoption 14.794 p < .001

Welch’s two-sample independent t-tests were then used to determine if data could be

considered MAR. That is, whether there was no statistically significant mean difference in one

variable (e.g., confidence) if an item in a different variable (e.g., decision/adoption) was

unanswered (see Appendix D). To evaluate this, marginal mean imputation was used to conduct

the t-tests; without imputation, the t-tests would provide information about the frequency of

missing data, as opposed to the relations between the missing items. Sidak family-wise error

correction was used to determine the family-wise α-values for multiple t-tests. Missingness on

decision/adoption items could not be considered to be MAR, while missingness on other

variables did appear to be MAR (see Appendix D). Because all data are not MAR, multiple

imputation, the most statistically robust method, was used to account for missing items

(Carpenter, Kenward, & Vansteelandt, 2006).

Multiple imputation replaces missing data with values imputed from the values of all

existing data for that individual given the values for that score that tend to co-occur with similar

values for other individuals. This process of imputation is conducted for all individuals and all

items; the resulting model is then re-analyzed and the process carried on iteratively. For this

analysis, multiple imputation was conducted using the Mice package in R, Version 3.2.3 (2013).

This package uses different models to estimate data for nominal, ordinal, and interval values; this

is an advantage because the GGNPS contains a variety of items including Likert-scales, “yes/no”

items, and a knowledge score. In addition, the Mice package imputes data beginning with the

variable that has the least amount of data missing, utilizing the most complete variables to fill in

data on less complete variables and thus using the maximal amount of information for each

imputation. Finally, the Mice package also uses random draws of data to mimic natural

variability in the sample.

Four imputations were conducted with 50 iterations for each imputation. The multiple

iterations allow the imputed data to try multiple variations until a “best-fit” scenario is found.

ANOVAs compared means of the four imputations. Results demonstrated that there were no

statistically significant differences between the imputations (see Table 4). This indicates that

with four different imputations the data sets were consistently similar, demonstrating increased

likelihood that the data was imputed based on the parameters of the collected data. Table 4

Sample Characteristics

The final sample included 6,861 participants. The majority were staff nurses (53.02%, n

= 3,638/6,861) whose primary role was direct patient care (61.01%, n = 4,186/6,861); 42.09% (n

= 2,888/6,861) reported spending 81-100% of their time seeing patients (see Table 6). The

GGNPS is intended to evaluate the use of genetics/genomics by nurses providing direct patient

care, therefore this sample is representative of the target population. The majority of participants

were female (71.05%, n = 4,875/6,861), White (62.26%, n = 4,272/6,861), and

nonHispanic/Latino (72.15%, n = 4,950/6,861) (see Table 5). A large portion (44.67%, n =

3,065/6,861) of the participants had a baccalaureate education (see Table 5). Among the

participants, 10.65% (n = 731/6,861) reported being current students (see Table 5). Participants

also reported their age (see Table 5); however, time since graduation or length of nursing career

cannot be inferred based on reported age because educational and career paths vary significantly.

The age of four participants was not reported because they incorrectly wrote 2010 or 2012 as

their birth year.

Social System

6861

Imputation: 2.33

1.50

.484

.693

Imputation: 2.33

1.50

Imputation: 2.31

1.49

Imputation: 2.34

1.50

Decision/Adoption

6861

Imputation:

1.13

2.16

.135

.939

Imputation:

1.12

2.16

Imputation:

1.14

2.15

Imputation:

1.12

2.17

Table 5

Frequency Table – Demographic Variables.

Variable

Frequency

Percent of

Participants

Percent of Participants with

Non-Responders Removed

Gender

Female

4875

71.05%

93.7%

Male

328

4.78%

6.3%

No Response

1658

24.17%

Ethnicity/Race

American Indian/Alaska Native

.38%

.51%

Asian

384

5.60%

7.6%

Black/African American

335

4.88%

6.63%

Native Hawaiian/Pacific Islander

.50%

.67%

White

4272

62.26%

84.58%

No Response

1810

26.38%

Hispanic/Latino

Yes

231

3.37%

4.46%

4950

72.15%

95.54%

No Response

1680

24.49%

Age

20-45 years

2354

34.31%

49.30%

46-65 years

2328

33.93%

48.76%

66+ years

1.34%

1.92%

No Response

2087

30.42%

Highest Nursing Degree

Associate Degree in nursing

1062

15.48%

20.36%

Baccalaureate Degree in nursing

3065

44.67%

58.77%

Diploma

324

4.72%

6.21%

Doctorate Degree in nursing

.48%

.63%

Master’s Degree in nursing

731

10.65%

14.01%

No Response

Are you a student?

1646

23.99%

4491

65.46%

86.00%

Yes

No Response

731

1639

10.65%

23.89%

14.00%

ANA Member

3963

57.76%

75.93%

Yes

1256

18.31%

24.06%

No Response

1642

23.93%

Table 6

Frequency Table – Clinical Experience

Variable

Frequency

Percent of

Participants

Percent of Participants with

Non-Responders Removed

Primary Area of Expertise

Case Manager

1.40%

1.93%

Clinical Nurse Specialist

110

1.60%

2.21%

Consultant

.69%

.94%

Director/Assistant Director

109

1.59%

2.19%

Educator

230

3.35%

4.62%

Head Nurse

286

4.17%

5.75%

Nurse Practitioner

185

2.70%

3.72%

Researcher

.60%

.82%

Staff Nurse

3638

53.02%

73.14%

Supervisor

232

3.38%

4.66%

No Response

1887

27.50%

Primary Function

Administration

464

6.76%

9.27%

Education

272

3.96%

5.44%

Patient care

4186

61.01%

83.67%

Research

1.05%

1.44%

Student

.13%

.18%

No Response

1858

27.08%

Percent Time Seeing Patients

0-20%

741

10.8%

14.82%

21-40%

256

3.73%

5.12%

41-60%

374

5.45%

7.48%

61-80%

741

10.8%

14.82%

81-100%

2888

42.1%

57.76%

No Response

1861

27.12%

Number of Years Working in

Nursing

0-3 years 1137 16.57% 21.98%

6-10 years 744 10.84% 14.38% 11-15 years 606 8.83% 11.71%

16-20 years 643 9.37% 12.43%

21-25 years 502 7.32% 9.70%

26-30 years 587 8.56% 11.35%

31-35 years 524 7.64% 10.13%

36-40 years 316 4.61% 6.11%

41-45 years 97 1.41% 1.88%

46-50 years 17 .25% .33%

No Response 1688 24.6%

Content Validity

All eight content experts completed and returned the CVI. One individual answered the

questions intended for the learners, not the content experts, and this reply was excluded. The

remaining seven responses were used to calculate the CVR using Lawshe’s formula (Lawshe,

1975):

CVR = [Ne – (N/2)]/ (N/2)

A relevance table with four categories (ranging from ‘very relevant’ to ‘not relevant’) was

used to score each item. ‘Ne’ was determined by the number of panelists who classified the item

as ‘very relevant’ or ‘relevant’ and ‘N’ is the total number of content experts. Higher CVR

values indicate a greater degree of content validity as reported by the content experts. A value of

1 indicates that all reviewers considered an item ‘very relevant’ or ‘relevant,’ lower numbers

indicate that some of the reviewers considered the items ‘somewhat relevant’ or ‘not relevant.’

The majority of the items performed well (CVR = -0.714 to 1), see Appendix E for the

CVR values of each item in the GGNPS. However, several items had lower CVR values (-0.714

to 0.428), indicating that the content experts considered these items less relevant to the

instrument. Written comments from the content experts provided additional information. These

items included:

Domain: Confidence

Item: Confidence regarding the following: “Give patients information about the

limitations of genetic testing for common diseases.”

CVR = 0.428

One content expert stated that not all RNs may be able to provide this information.

Instead RNs may be able to reinforce information provided by genetics specialists and recognize

the need for clarification.

Domain: Competency/Knowledge

Item: “Thinking about how you support clinical decisions (such as administering drugs

prescribed), how important do you think each of the following is to consider?”

CVR= 0.428

Two content experts indicated this item would need more specific information. For

example, a gene known to contribute to ovarian cancer such as BRCA1 and 2, as opposed to a

more multifactorial disease such as diabetes. The second expert added that it was unclear

whether the item was referring to medication administration or was a general question; also, that

a clinical example that required consideration for a referral may clarify the item.

Item: “The DNA sequences of two randomly selected healthy individuals of the same sex

are 90-95% identical.”

CVR= - 0.714

Regarding DNA sequence similarities among two randomly selected healthy individuals,

a content expert indicated that this item may not have clinical significance and the DNA

similarities may exceed 95% (as opposed to the 90-95% indicated by the GGNPS). However,

further analysis of this item is required because this is a ‘True/False’ item and the correct answer

is ‘False.’ At least one content expert evaluated this item with the assumption that the correct

answer was ‘True.’ As a result, additional clarification from content experts is warranted.

Item: “In describing your genetic/genomic knowledge, would you consider it to be”

CVR= 0.428

This item was considered redundant by a content expert because a previous item asks

learners to rate their understanding of the genetics of common diseases as excellent, good, or

poor.

Domain: Social Systems

Item: “Do you intend to learn more about genetics?”

CVR = 0.142

Two content experts stated that this item may overlap with the attitudes/receptivity

domain, as opposed to social systems. A third content expert stated that this item may be

misleading because she believes all participants will answer ‘yes.’

Lawshe’s method for determining content validity is widely used in a variety of fields,

including health sciences, psychology, and education. Despite its frequent use, it has some

limitations. Lawshe (1975) reported that the critical values for content validity were calculated

by Lowell Schipper. Neither Schipper nor Lawshe described the method for calculating the

values; when statisticians attempted to recreate the calculations using discrete binomials and

normal approximation to the binomial, neither exactly matched the values calculated by

Schipper. Wilson and colleagues (2012) hypothesized that variation may be due to a calculation

error, typographical error, or incorrect labeling by Lawshe. Lawshe reported critical CVR at α =

0.05, one tailed; however, the values calculated by Schipper are a closer match to a normal

approximation to the binomial at α = 0.05, two tailed (or α = 0.025, one tailed).

Lastly, Wilson and colleagues (2012) hypothesized that Schipper may have selected,

instead of calculating, that a critical CVR for n = 5, 6, 7 was 0.99, which may explain why it

does not match attempts to recalculate the values. Due to a small sample size of content experts,

he may have considered agreement among all reviewers mandatory to reach critical CVR.

However, when calculated by Wilson and colleagues (2012), using normal approximation to the

binomial, the critical CVR for n = 7 was 0.741 (p = 0.05, two tailed test). A CVR of 0.741 (n =

7) is achieved if only one content experts rates an item as ‘not relevant’ or ‘somewhat relevant.’

Therefore, both calculating the CVR using normal approximation to the binomial and

consideration that all, except one, content experts consider the item ‘relevant’ or ‘very relevant’

is the basis for considering items achieving a score of 0.741 or greater as relevant to the GGNPS.

A comprehensive CVI was computed as the mean of the content validity ratios across all

items. The range for agreement varied among content experts, -0.714 to 1 (see Appendix E). The

overall CVI was 0.805, exceeding 0.741, which Wilson and colleagues (2012) identified as the

required threshold when seven content experts evaluate an instrument. This information was also

used in the ancillary analysis in combination with results from the exploratory factor analysis

(EFA) to evaluate items considered for removal from the instrument.

Face Validity

Face validity was conducted with seven reviewers. All reviewers were nurses with a

variety of clinical and educational backgrounds. Educational backgrounds varied from associate

to doctoral degrees (see Table 7). Number of years in clinical practice varied from 2 to 31 years.

None of the reviewers had specific education or expertise in genetics/genomics. Reviewers

without genetics/genomics expertise were selected to provide the perspective of the majority of

nurses in the workforce. The GGNPS was created to evaluate the clinical applications of

genetics/genomics; therefore, reviewers who spend the majority of their time seeing patients

were selected. Face validity was conducted to evaluate the ease of understanding and applying

the instrument to nursing practice.

Most of the participants reported that the GGNPS was somewhat clear/easy to understand

(n = 5/7); others reported that it was very clear/easy to understand (n = 2/7) (see Table 7). The

majority of participants (n = 5/7) reported that the items were pertinent to their clinical practice

(see Table 7). Feedback from the reviewers indicated the GGNPS has face validity.

(3ng/poy)

Aseo/reo]3

area

osInN

Sox

yeyMowlos

,JO\Sey|

quoned

%OL

peoyH Or L

Aseo/1ea[o

aed

SOX Sox

yeyMowos

,JO]SeJ|

quoned

%OOI

OSINN

JJRIS

IZ 9

Jouon

Aseo/1eo[o

-oelg

Sox Sox

AINA,

Josey

uoneonpy

“Or

osINN

I¢

Asea/teayo

aed

Sox Sox

yeYMOUIOS

91210}90q

quoned

%OOT

OSINN

JJeIS

Aseo/teayo

(aa)

area

yeYMOUIOS

oyeomneypeooeg

quoned

%09

OSINN

JJeIS

€

Aseo/teayo

(NOD

2Re9

sox

yeYMOUIOS

oyeomneypeooeg

quoned

%OOT

OSINN

JJeIS

v Z

Aseo/1eo]o

ored

AIO.

a1e1D0SsV

quoned

%001

9SINN

FPS

0€

gsonoeid

sulsinu

[eorurys

ano

{puejsiopun

0} ul

ursormouss

«goonoevid

jeorurja

sea

reajo

sonjoeid

syuoryed

Roly

SuLyIOM

/sonouss

NoK

JuOUTVIOd

quouNsUI

seISop

Sursinu

jovoie

Suloosoum

yeuonouny

sieaX

asn

noAoq

suonsonb

ory

ou}

Ss]

ysoysty

Areulig

uonsjodolg

Aeulig

JOquUNN

JOMOIADY

‘sy[nsoy

Ape

90ey

LAqeL

Construct Validity

Hypothesis 1.

Confirmatory Factor Analysis results will show that:

a. GGNPS items Part 2, Questions 2.1, 2.2, 2.4; Part 4, Questions 1, 2, 3; Part 5,

Questions 1, 2; Part 6, Questions 1, 2 comprise the underlying latent variable of

COMPETENCY/ KNOWLEDGE.

b. GGNPS items Part 1, Questions, 1, 2, 3; Part 2, Questions 2, 3 comprise the

underlying latent variable of ATTITUDES/RECEPTIVITY.

c. GGNPS items Part 3, Questions 2, 3, 4 comprise the underlying latent variable of

DECISION/ADOPTION.

d. GGNPS item Part 2, Question 1 comprises the underlying latent variable of

CONFIDENCE.

e. GGNPS item Part 7, Questions 1.3, 1.4, 1.5, 1.6, and 1.7 comprise the underlying

latent variable of SOCIAL SYSTEM.

A CFA was conducted to evaluate whether the respective GGNPS items comprised the

underlying variables. The CFA was conducted with the original data set (with missingness

handled through listwise deletion in MPLUS) and with the four data sets imputed through the

Mice package in R. Conducting the analysis with five models, the original data and four

imputations, allowed comparison between versions of imputed data. Multiple imputations were

evaluated for consistency among the results. MPLUS, version 7.3, was used to compute both the

CFAs and structural equation models, conducted full information maximum likelihood

estimations of the missing data; this is the default setting for MPLUS to handle missing-at-random

or not-missing-at-random missing ordinal and nominal data (Muthén and Muthén, 2012) and is

the optimal method for these data. After, e.g., 34 hours and 38 minutes, and 460 iterations,

MPLUS reported that the models did not converge on a stable set of parameter estimates, as a

result no stable factor loadings can be reported.

Hypothesis 2.

Structural equation model (SEM) fit to data. Testing the SEM using item-level entities

required more computing power than could be achieved; therefore, the analyses were conducted

in two steps. The first step was the creation and evaluation of factor scores (by combining the

items associated with each variable); this was followed by a series of models to test the relation

between factors. SEMs are reported for the original data (see Figure 7) and the imputed data sets

(see Appendices F.1-F.4).

A SEM evaluated the proposed relations (see Figure 7). The proposed model did not fit

the collected data using χ-square as an initial fit test (see Table 8). The large χ-square (451.979, p

< .001) indicates a poor fit of the model to the data. However, because χ-square is often

significant in large sample sizes (Hooper, Coughlan, & Mullen, 2008), other measures of fit were

analyzed. Root mean square error of approximation (RMSEA) is an evaluation of fit that

considers sample size. An RMSEA ≤ 0.05 is typically regarded as indicating a very good fit and

≤ 0.08 is considered to indicate an acceptable fit (Browne et al., 1993). In the original data set

and the four variations of imputed data, none of the models indicated a good fit (see Table 8).

The Comparative Fit Index (CFI) compares a posed model with a null model and CFI indices

range between 0 and 1. Values ≥ 0.9 indicate a good model fit (Davis & Murrell, 1993). In the

original data set and the four variations of imputed data, none were greater than 0.9 (see Table 8).

Bayesian Information Criterion (BIC) is also a comparative fit measure between models that

corrects for the number of parameters in a model as well as the sample size (Raftery, 1993). A

low BIC value indicates a better model fit. Among the original data set and the four variations of

imputed data, the BICs ranged from 79,506.863-139,625.503 (see Table 8). Multiple fit

evaluations including χ-square, RMSEAs, CFIs, and BICs all indicated that the hypothesized

model did not fit the data. Furthermore, the poor fit was consistent across the original data set

and all four imputations.

Table 8

Model Fit Using the Original Data Set and Multiple Imputations.

Original Data Set

Imputed 1

Imputed 2

Imputed 3

Imputed 4

Chi-square

451.979

df= 3 p

<.001

603.970

df= 3 p

< .001

657.126

df= 3 p

< .001

617.126

df= 3 p

< .001

593.837

df= 3 p

< .001

RMSEA

.170

.160

.167

.162

.159

CFI

.839

.875

.871

.872

.877

BIC 79,506.863 139,302.730 139,205.026 139,625.503 139,297.560

R2 illustrates how much of the variance was explained by the model (see Table 9). This

includes the original data set and multiple imputations. Despite the fact that a statistically

significant amount of variance was explained by the model, the normalized residuals for the

covariances and correlations indicate a significant amount of residual variance was not addressed

(see Table 10).

Table 9

R2 square Values for Original Data and Multiple Imputations.

Original Data

Imputation 1

Imputation 2

Imputation 3

Imputation 4

Attitude/

Receptivity

R2 = .031 p

< .001

R2 = .029

p < .001

R2 = .040

p < .001

R2 = .030 p

< .001

R2 = .031

p < .001

Confidence

R2 = .202 p

< .001

R2 = .152

p < .001

R2 = .153

p < .001

R2 = .148 p

< .001

R2 = .146 p

< .001

Knowledge/

Competency

R2 = .166 p

< .001

R2 = .133

p < .001

R2 = .149

p < .001

R2 = .138 p

< .001

R2 = .135 p

< .001

Decision/

Adoption

R2 = .134 p

< .001

R2 = .184

p < .001

R2 = .180

p < .001

R2 = .179 p

< .001

R2 = .190 p

< .001

Table 10

Normalized Residuals for the Original Data and Multiple Imputations.

Original

Data

Imputation 1

Imputation 2

Imputation 3

Imputation 4

Decision/Adoption

and Confidence

16.50

16.76

17.01

16.80

16.56

Social systems and

Confidence

5.52

6.60

7.25

6.33

7.18

Knowledge/

Competency and

Decision/Adoption

12.19

15.85

16.45

16.15

15.4

Individual Variable Fit

Figure

SEM Using the Original Data Set.

p* < 0.001; p** = 0.013

Hypothesis 2a. COMPETENCY/KNOWLEDGE of genetics/genomics has a

statistically significant direct effect on one’s ATTITUDES/RECEPTIVITY and

CONFIDENCE. COMPETENCY/KNOWLEDGE has a statistically significant indirect

effect on one’s DECISION/ADOPTION of genetics/genomics.

In both the original and imputed data, there was a statistically significant direct effect of

competency/ knowledge on attitudes/receptivity (see Figure 7 and Appendices F.1-F.4).

However, considering the large sample size, the effect size was small, indicating a weak relation

between competency/ knowledge on attitudes/receptivity (.104). The original and imputed data

demonstrated a statistically significant direct effect of competency/ knowledge on confidence

(.399) (see Figure 7 and Appendices F.1-F.4). The indirect relation (-.023) between competency/

knowledge and decision/adoption (see Table 11) was negative and weak.

Social Systems

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

.391*

0.150*

0.717*

0.399

0.039**

0.104*

0.225*

Hypothesis 2b. ATTITUDES/RECEPTIVITY has a statistically significant direct

effect on one’s DECISION/ADOPTION of genetics/genomics.

The relation between attitudes/receptivity on decision/adoption was -.225 (see Figure 7),

the relation for imputed data is described in Appendices F.1-F.4).

Hypothesis 2c. SOCIAL SYSTEMS has a statistically significant direct effect on

COMPETENCY/KNOWLEDGE, ATTITUDES/RECEPTIVITY, and

DECISION/ADOPTION and a statistically significantly indirect effect on one’s

ATTITUDES/RECEPTIVITY and DECISION/ADOPTION of genetics/genomics.

The relation between social systems and competency/ knowledge (.717) was statistically

significant and the strong (see Figure 7 and Appendices F.1-F.4). By contrast, the relation

between social systems and attitudes/receptivity (.150) was weak, although statistically

significant (see Figure 7 and Appendices F.1- F.4). The indirect relation between social systems

and decision/adoption (.034) was a negative and weak relation (see Table 11).

Hypothesis 2d. CONFIDENCE has a statistically significant direct effect on

ATTITUDES/RECEPTIVITY, and a statistically significant indirect effect on

DECISION/ADOPTION of genetics/genomics.

The direct relation between confidence and attitudes/receptivity (.039) was the weakest

relation in the SEM and was not statistically significant in imputed data set 1 (see Figure 7 and

Appendices F.1-F.4). The indirect relation between confidence and decision/adoption (.0008) the

weakest indirect relation in the SEM (see Table 11).

Table 11

Indirect Path Model Relations.

Original

Data

Imputation

Imputation 3

Imputation

Competency/

Knowledge &

Decision/Adoption

-.023

-.020

-.030

-.023

-.026

Social Systems

&Attitudes/Receptivity

.075

.039

.061

.046

.050

Social Systems &

Decision/Adoption

-.034

-.084

-.091

-.083

-.082

Confidence &

Decision/Adoption

.0088

.0084

.011

.015

.012

Ancillary Analysis

Expert review supported the content validity and evaluation by a sample of RNs

supported face validity of the GGNPS. However, construct validity could not be established.

The CFA did not converge on a stable set of parameter estimates and reliable factor loadings

could therefore not be estimated. Similarly, the overall SEM model did not fit the data well and

many proposed relations, though statistically significant, were weak given the large sample size

and thus large power.

Due to these conflicting results between content and construct validity, a data driven

exploratory factor analysis (EFA) was conducted to compare factor loadings to the five proposed

variables in the GGNPS. Thus, a five-factor model was requested. Oblique rotation was used to

allow for factors to be related to each other (Osborne, 2015).

Criteria used for evaluation were factor loadings ≥.30 (Costello & Osborne, 2005) or

higher and presence of Thurstone’s (1947) simple structure. Thurstone’s simple structure

requires that 1) each row in a rotated factor matrix contains at least one loading near zero, 2)

among the columns there should be factors with zero loadings and other factors with higher

loadings, 3) variables should load on one factor (Plichta, Kelvin, & Munro, 2013; Thurstone,

1947).

Results

The EFA results for originally designated items (e.g., confidence, attitudes/receptivity

etc.) indicate that a). confidence items consistently load onto Factor 1, b). attitudes/receptivity

items and items from competency/ knowledge that are related to nurses’ attitudes regarding

genetics/genomics load onto Factor 2, c). decision/adoption items and items from

attitudes/receptivity that are related to perceived disadvantages of adopting genetics/genomics

into nursing practice load onto Factor 3, d). many competency/ knowledge items load onto Factor

4, specifically items related to collection of family history information, e). competency/

knowledge items also load onto Factor 5, specifically items related to clinical decision making

(see Table 12). Notably, social system items did not consistently load onto any single factor,

some items loaded on two other factors, while other items did not load above .30.

Table 12

Factor Loading Overview.

GGNPS Variable

Factor 1

Confidence

Factor 2

Attitudes/receptivity

Factor 3

Decision/adoption

Factor 4

Competency/knowledge 1

Factor 5 Competency/knowledge 2

Factor 1: Of the 14 GGNPS-designated confidence items (Calzone et al., 2014), all but

one item met both the loading criteria (≥ .30) and demonstrated Thurstone’s simple structure (see

Table 13). Item P2_Q1_05, asking nurses to rate their confidence in providing emotional

followup support after family history assessment, loaded most strongly on Factor 1 (.670) it also

loaded above .30 (.308) on Factor 2 (attitudes/receptivity).

Table 13

EFA Results for GGNPS-Designated Confidence Items.

Item

P2_Q1_01

Decide what family history information is

needed to tell something about a patient’s genetic

susceptibility to common diseases.”

.751*

.115

.232

.281

-.190

P2_Q1_02

Use family history information to support

treatment decisions for patients who have

common diseases.”

.742*

.138

.232

.288

-.180

P2_Q1_03

Counsel patients about inherited risk for

common diseases.”

.758*

.105

.183

.194

-.127

P2_Q1_04

Assure that patient’s family history information

relating to common diseases remains

confidential.”

.383*

.283

-.009

.015

-.133

P2_Q1_05

Provide emotional follow-up support after family

history assessment.”

.670*

.308

-.116

-.035

-.124

P2_Q1_06

Discuss how family history affects recommended

screening intervals.”

.768*

.106

.096

.132

-.080

P2_Q1_07

Decide which patients would benefit from a

referral for genetic counseling and possible

testing for susceptibility to common diseases.”

.817*

.019

.026

.115

.002

P2_Q1_08

Access reliable and current information about

genetics and common diseases.”

.755*

.018

-.036

.057

.023

P2_Q1_09

Provide information about the availability of

genetic testing for common diseases.”

.882*

-.093

-.053

-.050

.145

P2_Q1_10

Give patients information about the risks of

genetic testing for common diseases.”

.925*

-.105

-.017

-.073

.173

P2_Q1_11

Give patients information about the benefits of

genetic testing for common diseases.”

.925*

-.074

-.008

-.074

.162

P2_Q1_12

Give patients information about the limitations

of genetic testing for common diseases.”

.931*

-.109

-.013

-.081

.180

P2_Q1_13

Facilitate referrals for genetic services for common

diseases.”

.817*

-.052

-.093

-.006

.123

P2_Q1_14

Provide emotional follow-up support after genetic

testing.”

.716*

.220

-.149

-.068

-.044

* Highest loading. Bolded items loaded ³ .30.

Factor 2: Of the 17 GGNPS-designated attitude/receptivity items (Calzone et al., 2014),

six loaded onto Factor 2 (see Table 14). These were items related to the potential advantages of

genetics/genomics. Eight items of the 17 GGNPS-designated attitude/receptivity items loaded on

Factor 3. These items related to potential disadvantages of genetics/genomics and are discussed

below. Concerning the remaining three items, item P1_Q3_06 (would increase insurance

discrimination) was excluded from the EFA because the item did not demonstrate any variation

and thus provided no information about associations. All participants, except those who left the

item blank, responded “yes” indicating this was a perceived disadvantage of genetics/genomics.

The remaining two items did not load above .30 on any factor and did not demonstrate

Thurstone’s simple structure. These items were, P1_Q1 (importance of nurses to become

educated about genetics of common diseases) and P1_Q3_04 (if a potential disadvantage of

integrating genetics is that it would not improve patient care). Item P2_Q3_03 (family history

taking should be a key component of nursing care) loaded ³ .30 on Factor 4

(competency/knowledge 1) as well as .375 on Factor 2 (attitudes/receptivity). The other items

demonstrated Thurstone’s simple structure. Eight, out of 25, GGNPS-designated

competency/knowledge items loaded onto Factor 2 (see Table 16). However, these were items

can be considered related to nurses’ attitudes regarding genetics/genomics. For example, asking

nurses if they think genetic risk is associated with a variety of clinical conditions.

Table 14

EFA Results for GGNPS-Designated Attitudes/Receptivity Items.

Item

P1_Q1

How important do you think it is for the nurse to

become more educated about the genetics of

common diseases?”

.155

.282*

-.058

.198

.151

P1_Q2_01

Potential advantages of integrating genetics: Better

decisions about recommendations for

preventative services.”

.042

.377*

.206

.153

.106

P1_Q2_02

Potential advantages of integrating genetics: Better

treatment decisions.”

.053

.314*

.190

.070

.122

P1_Q2_03

Potential advantages of integrating genetics:

Improved services to the patients.”

-.020

.306*

.094

.077

.120

P1_Q2_04

Potential advantages of integrating genetics: Better

adherence to clinical recommendations among

patients.”

.013

.343*

.218

.114

.133

P1_Q2_05

Potential advantages of integrating genetics:

Genetic risk triaging could make better use of

visit time.”

.064

.325*

.186

.118

.151

P1_Q3_01

Potential disadvantages of integrating genetics:

Would take too much time.”

-.045

-.124

.369*

-.022

.063

P1_Q3_02

Potential disadvantages of integrating genetics:

Not reimbursable/Too costly.”

.040

.028

.397*

.070

.045

P1_Q3_03

Potential disadvantages of integrating genetics:

Need to ‘re-tool’ professionally.”

-.027

.013

.321*

.018

.057

P1_Q3_04

Potential disadvantages of integrating genetics:

Would not improve patient care.”

-.004

-.292*

.157

-.118

-.120

P1_Q3_05

Potential disadvantages of integrating genetics:

Increase patient anxiety about risk.”

-.046

.013

.330*

.072

-.03

P1_Q3_07

Potential disadvantages of integrating genetics:

Medical-legal problems for nurses related to

testing.”

-.014

-.005

.430*

.038

.057

P1_Q3_08

Potential disadvantages of integrating genetics:

Greater burden of responsibilities on nurses.”

-.067

-.051

.448*

-.076

.042

P1_Q3_09

Potential disadvantages of integrating genetics:

Need to educate nurses in genetics.”

-.142

.059

.370*

.032

.031

P1_Q3_10

No disadvantages.”

.110

.121

-.657*

-.071

.009

P2_Q3_03

Family history taking should be a key component

of nursing care.”

.072

.375*

-.144

.304

.028

*Item P1_Q3_06 “Would increase insurance discrimination” is not included because there was

no variation in the responders’ reply.

Factor 3: Analysis of the EFA results indicates Factor 3 relates to decision/adoption. Of

the 12 GGNPS-designated decision/adoption items, eight loaded onto Factor 3 (see Table 15).

Those items also demonstrated Thurstone’s simple structure. Of the remaining four items, one

item, P3_Q3 (in the past 3 months, has any patient initiated a discussion with you about

genetics), did not load above .30 on any factor (see Table 15) and three items designated as

decision/adoption in the GGNPS loaded onto Factor 4 (competency/knowledge 1) and are

discussed below. As previously noted in the Factor 2 section, eight of the 17 GGNPS-designated

attitude/receptivity items, loaded onto Factor 3 (decision/adoption), see Table 14. These items

related to potential disadvantages, and therefore also perhaps related to the decision to adopt

genetics and genomics.

Table 15

EFA Results for GGNPS-Designated Decision/Adoption Items.

Item

P2_Q2_01

Limitations of discussing genetics with

individuals “My difficulty finding information on

genetics and common diseases.”

-.070

.065

.448*

-.021

.076

P2_Q2_02

Limitations of discussing genetics with

individuals “Lack of time to get a patient’s

complete family history.”

.058

-.038

.523*

-.085

.081

P2_Q2_03

Limitations of discussing genetics with

individuals “No reimbursement for counseling or

testing.”

.143

-.025

.564*

.077

.097

P2_Q2_04

Limitations of discussing genetics with

individuals “No place to send patients for

counseling or testing.”

.072

.069

.582*

-.001

.052

P2_Q2_05

Limitations of discussing genetics with

individuals “Increased nursing liability.”

.073

-.091

.606*

-.075

.114

P2_Q2_06

Limitations of discussing genetics with

individuals “Not in my scope of practice.”

-.140

-.006

.363*

-.153

-.009

P2_Q2_07

Limitations of discussing genetics with

individuals “My limited expertise in genetics.”

-.297

.291

.435*

-.057

-.006

P2_Q2_08

Limitations of discussing genetics with

individuals “Lack of use of genetics in my

clinical area.”

-.146

.184

.446*

-.121

-.004

P3_Q2

In the past 3 months, how often have you collected a

complete family history …”

.191

-.057

-.040

.401*

.099

P3_Q3

In the past 3 months, has any patient initiated a

discussion with you about genetics?”

.182

-.006

-.182

.297*

.023

P3_Q4_01

In the past 3 months, how often have you used

family history information when facilitating

clinical decisions or recommendations for your

patients?”

.197

.053

-.065

.402*

.044

P3_Q4_02

In the past 3 months, how often have you facilitated

referrals to genetic services?”

.264

-.139

-.246

.318*

.099

Factor 4: Analysis of the EFA results indicates Factor 4 relates to

competency/knowledge. However, of the 25 GGNPS-designated competency/knowledge items

(Calzone et al., 2014), only seven loaded on Factor 4 (see Table 16). These were items related to

collection of family history information. As previously noted, 8 GGNPS competency/knowledge

items loaded on Factor 2 (attitudes/receptivity) and as noted below six competency/knowledge

items loaded on Factor 5(competency/knowledge 2). The remaining four GGNPS-designated

competency/knowledge items did not have a loading above .30 or did not have simple structure.

These included items P2_Q3_01, P5_Q1, P5_Q2, and P6_Q1 (see Table 16). These items were

related to family history collection for new patients, knowledge of the Essential Competencies

and Curricula Guidelines for Nurses in Genetics and Genomics, and information regarding the

structure and function of genetic material.

In addition to competency/knowledge designated items, one of the five

GGNPSdesignated social system items loaded on Factor 4; item P7_Q1_01 (“do you intend to

learn more about genetics”). Finally, Factor 4 also includes three GGNPS-designated

decision/adoption items, P3_Q2, P3_Q4_01, P3_Q4_02 (see Table 16). These three items

explore nursing actions in the past three months related to collection and use of family history

information, as well as facilitating referrals to genetic services.

Factor 5: The EFA results indicate that Factor 5 also relates to competency/knowledge.

Of the 25 GGNPS-designated competency/knowledge items (Calzone et al., 2014), six loaded on

Factor 5. These were items related to clinical decision making.

Table 16

EFA Results for GGNPS-Designated Competency/Knowledge Items.

Item

P2_Q3_01

A family history that includes only 1st degree

relatives such as parents, siblings, and children

should be taken for every new patient.”

-.076

-.173*

-.027

-.056

.002

P2_Q3_02

A family history that includes 2nd and 3rd degree

relatives such as grandparents, aunts, uncles, and

cousins should be taken for every new patient.”

.078

0.307*

-.111

.252

.021

P2_Q3_04

There is a role for nurses in counseling patients

about genetic risks.”

.144

.339*

-.166

.232

.045

P4_Q1_01

Do you think genetic risk has clinical relevance

for: Breast cancer.”

-.020

.923*

-.002

.011

.016

P4_Q1_02

Do you think genetic risk has clinical relevance

for: Colon cancer.”

.001

.880*

.002

.001

.013

P4_Q1_03

Do you think genetic risk has clinical relevance

for: Coronary heart disease.”

-.033

.826*

-.013

.007

.028

P4_Q1_04

Do you think genetic risk has clinical relevance

for: Diabetes.”

-.018

.786*

-.007

-.011

.037

P4_Q1_05

Do you think genetic risk has clinical relevance

for: Lung cancer.”

.017

.646*

-.011

-.076

.139

P4_Q1_06

Do you think genetic risk has clinical relevance

for: Ovarian cancer.”

.016

.778*

-.025

-.032

.066

P4_Q2_01

When patients indicate a disorder in the

family, which information do you collect in

your standard family history assessment:

“Age at diagnosis of condition”

.022

-.129

-.023

.836*

.007

P4_Q2_02

When patients indicate a disorder in the

family…: “Relationship to the patient.”

-.159

.044

.029

1.019*

-.204

P4_Q2_03

When patients indicate a disorder in the

family…: “Race or ethnic background.”

.047

-.085

.004

.660*

.101

P4_Q2_04

When patients indicate a disorder in the

family…: “Age at death from condition.”

-.020

-.147

.006

.900*

-.029

P4_Q2_05

When patients indicate a disorder in the

family…: “Both sides of the family

(maternal/paternal).”

-.111

.005

-.013

.951*

-.150

P4_Q4_01

Which of the following is important to

consider in clinical decision-making:

“Gender.”

-.019

.063

.045

.885*

P4_Q4_02

Which of the following is important to

consider in clinical decision-making:

“Race/Ethnicity.”

-.008

.060

.050

.008

.912*

P4_Q4_03

Which of the following is important to

consider in clinical decision-making:

“Genes.”

.012

.087

-.020

.086

.851*

P4_Q4_04

Which of the following is important to

consider in clinical decision-making: “Family

history.”

-.004

.127

-.003

.108

.820*

P4_Q4_05

Which of the following is important to

consider in clinical decision-making: “Age.”

-.028

.175

.008

.078

.496*

P4_Q4_06

Which of the following is important to

consider in clinical decision-making:

.059

-.022

.030

.130

.328*

“Insurance Status.”

P5_Q1

The DNA sequences of two randomly selected

healthy individuals of the same sex are 90-95%

identical.”

.118*

-.101

.041

.111

.033

P5_Q2

Most common diseases such as diabetes and

heart disease are caused by a single gene

variant.”

.102

-.092

.014

.132*

.016

P6_Q1

The Essential Competencies and Curricula

Guidelines for Nurses in Genetics and

Genomics are endorsed as being a standard

part of nursing practice. Have you heard or

read about these Competencies?”

.199

-.150

-.108

.281*

.154

P6_Q2_01

Please rate your understanding of the genetics

of common diseases.”

.445

-.354

.010

.496*

.119*

P6_Q2_02

In describing your genetic/genomic knowledge,

would you consider it to be.”

.445

-.375

-.003

.491*

.139

The GGNPS-designated social system items did not load predominantly onto one factor

(see Table 17). Item P7_Q1_01 loaded onto Factor 4. Two items, P7_Q1_02 (would you be able

to attend a course during work hours) and P7_Q1_03 (would you attend a course on your own

time) did not contain a loading above .30 and did not contain simple structure. Items P7_Q1_04

(do you think your senior staff members see genetics as an important part of your role) and

P7_Q1_05 (do you think your senior staff members see genetics as an important part of their

role) loaded ³ .30 on both Factors 3 (decision/adoption) and 4 (competency/knowledge).

Furthermore, both items P7_Q1_04 and P7_Q1_05 included negative loadings (see Table 17). A

negative factor loading indicates an inverse relation between the item and the factor. Social

systems may be pervasive throughout the other variables, as opposed to being an independent

variable.

Table 17

EFA Results for GGNPS-Designated Social System Items.

Item

P7_Q1_01

Do you intend to learn more about genetics?”

.059

.198

-.173

.308*

.137

P7_Q1_02

Would you be able to attend a course during work

hours?”

.027

.108

-.130

.116*

.114

P7_Q1_03

Would you attend a course on your own time?”

.027

.198

-.142

.215*

.091

P7_Q1_04

Do you think your senior staff members see genetics

as an important part of your role?”

.001

.082

-.633*

.387

.160

P7_Q1_05

Do you think your senior staff members see genetics

as an important part of their role?”

-.024

.092

-.635*

.370

.141

Factor loadings and simple structure results were also evaluated with CVRs, analyzing

EFA data with content validity expert input (see Table 18). Many items with loadings below .30

had high CVRs. In addition, these items contained important content, such as how to collect

family history information. Since they did not load above .30 in the EFA, they would be

recommended for deletion based on Thurstone’s simple structure; however, in consideration with

evaluation by the content experts they are recommended for revision. By contrast, item P5_Q1

had a low factor loading, lack of simple structure, and low CVR, it is recommended for deletion

or revision (see Table 18).

Table 18

Items with Low Factor Loadings and Associated CVRs.

Item

Highest

Simple

Structure

CVR

P1_Q1

“How important is it for the nurse to become more

educated about the genetics of common diseases?”

.282

Absent

P1_Q3_04

“Potential disadvantages: would not improve patient

care.”

.292

Absent

N/A

P2_Q3_01

“A family history that includes only 1st degree relatives

such as… should be taken for every new patient.”

(shortened)

-.173

Absent

.714

P5_Q1

“The DNA sequences of two randomly selected healthy

individuals of the same sex are 90-95% identical.”

.118

Absent

-.714

P5_Q2

“Most common diseases such diabetes and heart

disease are caused by a single gene variant”

.132

Absent

P6_Q1

“Have you heard of the Essential Competencies… in

Genetics and Genomics?” (shortened)

.281

Absent

0.714

P7_Q1_02

“Would you be able to attend a course during work

hours?”

.116

Absent

P7_Q1_03

“Would you be able to attend a course on your own

time?”

.215

Absent

.714

Analyses support that the GGNPS does have content and face validity. However,

statistical analyses also indicate that the constructs in the GGNPS may be more interrelated than

originally hypothesized. A number of items from competency/knowledge loaded on

attitudes/receptivity. In addition, attitudes/receptivity items related to perceived disadvantages

loaded onto decision/adoption. Some social systems items did not load well on a single factor,

while others loaded on multiple factors simultaneously. This may indicate that social systems

pervade throughout the other variables, as opposed to being an independent variable. Results of

the EFA also provide insight into why the CFA was unable to load on a stable set of parameters.

Many items loaded on factors different than those originally designated in the GGNPS. Further

analysis of how items in the GGNPS are related to the underlying variables can support future

analysis of construct validity.

Chapter V

Discussion

This study proposed to test the face, content, and construct validity of the GGNPS

developed by Calzone and colleagues (2012, 2016, 2014). The GGNPS is based on Rogers’

(2003) DOI theoretical framework and model, as well as Calzone and colleagues’ (2012) work,

which added confidence as a variable in the adoption of genetics/genomics into nursing practice.

Construct validity was explored in two ways, CFA and SEM analyses. The SEM was based on

the current study’s hypothesized model (see Figure 8), which stemmed from the works of Rogers

(2003) and Calzone and colleagues (2012; 2007).

Figure 7

Hypothesized Model.

Results indicate that the GGNPS has both content and face validity. However, construct

validity was not established. The CFA did not converge on a stable set of parameter estimates

and thus factor loadings could not be estimated. In addition, the SEM showed that this study’s

hypothesized model did not fit the data. However, the SEM did provide insight into the relations

between/amongst the model’s variables, though many of these relations were weak considering

the large sample size.

Social

Systems

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

Figure 8

Rogers’ DOI Model.

A number of factors may have contributed to the inability to establish construct validity.

First, the hypothesized model was based on Rogers’ DOI model (see Figure 9); however, the

model was revised for this study. Revisions included adding confidence as a variable, based on

work by Calzone and colleagues (2012), and removing portions of the model that were not

evaluated by the GGNPS. Comparing Figures 8 and 9, note that social system was the only prior

condition included in this study. Also, communication channels, characteristics of a

decisionmaking unit, and perceived characteristics of the innovation were not included in this

study. As previously mentioned, this study also did not evaluate implementation or confirmation,

instead focusing on adoption of genetics/genomics as per the work of Calzone and colleagues

(2014).

The second factor potentially affecting the evaluation of construct validity is the

designation of some GGNPS items as the variables: confidence, competency/knowledge,

attitudes/receptivity, decision/adoption, and social systems. The EFA, discussed below,

demonstrated that a number of items originally intended to measure one variable were actually

more highly related to a different variable; for example, several items thought to be part of the

competency/knowledge variable were more closely related to the attitudes/receptivity variable.

The third and last factor to consider is that the content validity results indicated that some items

may not fit the concepts being evaluated and may be considered for revision or removal. The

content validity experts scored several items lowly and some of these items also performed

poorly on the EFA, highlighting the need to revise or remove them. Because EFA results shed

light on the main analyses, the results of the ancillary analysis are discussed first.

Ancillary Data Findings

The EFA was performed with five factors requested because the GGNPS contains five

variables: confidence, attitudes/receptivity, decision/adoption, competency/knowledge, and social

systems. Many items did not load on their GGNPS-designated variables. Confidence was the

only variable where items loaded as predicted. However, attitudes/receptivity, decision/adoption,

and competency/knowledge were identifiable variables, though some of items loaded

unexpectedly on other variables. Also, while the GGNPS identified competency/knowledge as

one variable, the EFA results indicated that these items loaded on two factors, both with the

central theme of competency/knowledge. In contrast, social systems items did not load on a

factor unique to social systems; some items did not load on any factors, some loaded on multiple

factors, and one item loaded on competency/knowledge.

Factor 1. (Confidence).

Anticipated loadings. All of the 14 items designated as confidence in the GGNPS loaded

onto Factor 1. All but one item demonstrated Thurstone’s simple structure and met loading

criteria (³ .30) (see Table 13). In fact, most items loaded above .70, with three items loading

above .90, demonstrating a strong relation of these items to the confidence variable. The one

exception was item P2_Q1_05 which asked nurses to rate their confidence in providing

emotional follow-up support after family history assessment. This item did load most strongly

on Factor 1 (.670) but also loaded .308 on Factor 2 (attitudes/receptivity).

As noted previously, Rogers (2003) did not include confidence in the DOI theory, but

Calzone and colleagues (2012) added confidence items to the GGNPS, an instrument created

based on the DOI theory, because they found that confidence plays an important role in the

adoption of genetics/genomics into nursing practice. This has been demonstrated empirically:

Calzone and colleagues (2012) reported that only 24% of nurses (n =38/159) were very or

somewhat confident when deciding what family history information is required to assess

susceptibility to common diseases. Similarly, Metcalfe and colleagues (2009) reported that only

3% (n = 3/100) of hospice nurses were confident enough to provide client education and only

20% (n = 20/100) were very confident in explaining the genetic basis of transmission for

Duchenne Muscular Dystrophy. Both articles noted that low confidence may adversely affect

adoption of genetics/genomics.

In a similar vein, Benner (1984) discussed confidence related to gaining clinical

expertise. Her research on confidence helped define the confidence variable related to nursing

practice. Benner reported that confidence increases the ability to manage multiple roles and

challenges. Calzone and Jenkins (2007) identified a number of roles and responsibilities for

nurses in adoption of genetics/genomics. The 14 GGNPS confidence items ask participants to

report their confidence in relation to use of family history information, patient education,

providing emotional support, discussing confidentiality of family history information, and

facilitating referrals for genetic services.

Unanticipated loadings. Two GGNPS-designated competency/knowledge items also

loaded ³ .30 on Factor 1 (confidence). The first item, P6_Q2_01: “Please rate your

understanding of the genetics of common diseases” (participants rated their knowledge as

excellent to poor) loaded .496 on Factor 4 (competency/knowledge 1); however, it also loaded

.445 on Factor 1 (confidence). This item may have loaded on two factors because participants

have to both consider how much they know about genetics/genomics and how they would rate

their level of knowledge. The second item P6_Q2_02, parallels the previous one: “In describing

your genetic/genomic knowledge, would you consider it to be...”. This item loaded .491 on

Factor 4 (competency/knowledge 1). It also loaded .445 on Factor 1 (confidence). Again,

participants may have considered both their knowledge and their confidence in their knowledge.

Factor 2. (Attitudes/Receptivity).

Anticipated loadings. Of the 17 GGNPS-designated attitude/receptivity items, only six

loaded on Factor 2 (attitudes/receptivity) (see Table 14). These six items pertained to potential

advantages of genetics/genomics. While these six items did demonstrate Thurstone’s simple

structure and loaded ³ .30, the loadings ranged from .306-.377. This indicates a relatively weak

relation of the items to the attitudes/receptivity variable.

Concerning the other 11 items, two items did not load above .30 on any factor and one

item was excluded from the EFA because it did not contain any variability, all participants

responded with the same answer. The remaining eight items designated as attitudes/receptivity

loaded on Factor 3 (decision/adoption). Unlike the six attitudes/receptivity items that pertained

to advantages, these eight items pertained to potential disadvantages of incorporating

genetics/genomics into nursing practice and are discussed in detail in Factor 3.

Unanticipated loadings. Unexpectedly, eight GGNPS-designated

competency/knowledge items also loaded onto Factor 2 (attitudes/receptivity). These items may

be more strongly related to attitudes/receptivity than competency/knowledge because they asked

participants if they thought there was genetic risk associated with specific clinical conditions

(e.g. colon cancer) and if nurses thought family history information should be collected for new

patients. Rogers (2003) describes the formation of an attitude as participants processing their

thoughts about an innovation and forming a favorable or unfavorable opinion regarding the

innovation.

Factor 3. (Decision/Adoption).

Anticipated loadings. Of the 12 GGNPS-designated decision/adoption items, eight

loaded onto Factor 3 (decision/adoption) (see Table 15). These eight items pertained to

limitations of using genetics/genomics, such as lack of time, lack of reimbursement, increased

liability, and limited expertise. The eight items loaded ³ .30 and demonstrated Thurstone’s

simple structure. The loadings varied between .363 - .606, indicating a moderately strong

relation to the decision/adoption variable.

Of the remaining four items, one did not load above .30 on any factor and three loaded on

Factor 4 (competency/knowledge 1). Those three items pertained to facilitating referrals to

genetic services and family history information; the items loaded ³ .30 and demonstrated

Thurstone’s simple structure for the competency/knowledge factor.

Unanticipated loadings. As noted above, eight GGNPS-designated attitude/receptivity

(Factor 2) items unexpectedly loaded on Factor 3 (decision/adoption). The items loaded ³ .30

and demonstrated Thurstone’s simple structure for the decision/adoption factor. These items may

be more strongly related to decision/adoption than attitudes/receptivity because they concern

potential disadvantages of incorporating genetics/genomics into nursing practice. The content of

the GGNPS asked about potential disadvantages of genetics/genomics such as: taking too much

time, lack of reimbursement, increased burden of responsibilities on nurses, and increased patient

anxiety. Moreover, these particular previously designated attitudes/receptivity items parallel the

items in the GGNPS that are designated as decision/adoption.

That is, items in Factor 3 parallel Rogers’ description of decision/adoption because

individuals are considering potential disadvantages and limitations of adopting an innovation;

they are considering whether it is the best course of action. This is further supported by Rogers’

(2003) definition of decision and adoption. Rogers defines decision as the thought process

leading up to choosing to adopt or reject an innovation. Adoption is use of an innovation because

it is “the best course of action available” (Rogers, 2003, p. 177).

Decision/adoption is an important variable to explore in the EFA because it is the ultimate

outcome variable in the hypothesized model and it is affected by a number of the other variables.

For example, Calzone and colleagues (2014) reported that genetics/genomics are not being fully

adopted into nursing practice. The authors cited several reasons for lack of adoption including

inadequate knowledge, a lack of understanding of the relevance, lack of confidence to translate

knowledge into adoption of genetics and genomics, and inadequate support from social

system(s).

Factors 4 and 5 (Competency/Knowledge 1 and 2). There were 25 total GGNPS items

designated by Calzone and colleagues (2014) as competency/knowledge. While the GGNPS

identified competency/knowledge as a single variable, both Factors 4 and 5 were identifiable as

competency/knowledge factors and are thus designated as competency/knowledge 1 and 2. This

indicates that more than one type of genetics/genomics knowledge may affect adoption.

Similarly, Rogers (2003) describes three different types of knowledge. Awareness-knowledge is

knowing an innovation exists. This is basic knowledge that may encourage individuals to gain

more insight into an innovation. How-to-knowledge is learning how to use an innovation.

Lastly, principles-knowledge is understanding how an innovation functions.

This interpretation is further supported by the fact that, as previously mentioned, eight of

the 25 GGNPS-designated competency/knowledge items loaded onto Factor 2

(attitudes/receptivity). These items do not pertain to any of the three types of knowledge defined

by Rogers. Rather, they ask if nurses thought there was genetic risk associated with specific

clinical conditions and if nurses thought family history information should be collected for new

patients. The focus on participants’ perceptions, what they thought, rather than type of

knowledge supports the items loading on Factor 2 (attitudes/receptivity). This left 17 items

expected to load on Factors 4 and 5.

Factor 4 (Competency/Knowledge 1).

Anticipated loadings. Of the 17 remaining GGNPS-designated competency/knowledge

items, seven loaded on Factor 4 (see Table 16). These items related to collection of family

history information. All seven items loaded ³ .30 and demonstrated Thurstone’s simple structure.

The loadings’ strength varied widely, ranging from .491-1.019. This may be related to the

variation among competency/knowledge questions; these seven items cover a variety of clinical

topics, including both understanding of genetics/genomics and how it is used in clinical practice.

Factor 4 items parallel Rogers’ (2003) notion of how-to-knowledge: they are related to collection

and use of family history information, using this information in clinical decision making, and

facilitating referrals for genetic services. This type of knowledge is essential to understanding

how nurses use genetics/genomics in clinical practice. Identifying these items using the EFA can

allow educators to create targeted learning initiatives to teach topics that are required to

understand and apply genetics/genomics in a clinical setting.

Unanticipated loadings. As previously discussed, two GGNPS-designated

competency/knowledge items that loaded on Factor 4 also loaded ³ .30 on Factor 1 (confidence).

Item P6_Q2_01 loaded .496 on Factor 4 (competency/knowledge 1) and .445 on Factor 1

(confidence). Item P6_Q2_02 loaded .491 on Factor 4 (competency/knowledge 1) and .445 on

Factor 1 (confidence).

In addition, three items pertaining to facilitating referrals to genetic services and family

history information loaded on Factor 4 (competency/knowledge 1) (.318-.401) even though they

were originally designated as Factor 3 (decision/adoption). This may be because the items

related to participants’ how-to- knowledge about how to collect and utilize family history

information. Lastly, one GGNPS-designated social system item also loaded onto Factor 4

(competency/knowledge 1) (.308); the item asked whether participants intended to learn more

about genetics. Perhaps participants interpreted this item as intent to gain more how-to

knowledge and thus it loaded on competency/knowledge.

Factor 5. (Competency/Knowledge 2).

Anticipated loadings. Out of 10 remaining GGNPS-designated competency/knowledge

items, six loaded on Factor 5; these items were related to clinical decision-making (see Table 16).

All six items loaded ³ .30 and demonstrated Thurstone’s simple structure. Similarly to Factor 4,

loadings varied significantly for Factor 5, ranging from .328-.912. The items from this factor

parallel principles-knowledge. Principles knowledge is not required to adopt an innovation,

according to Rogers (2003), but promotes understanding of how an innovation functions. Factor

5 items include knowing whether gender, race/ethnicity, genes, family history, and insurance

status are important to consider in clinical decision-making. Although Rogers theorizes that

principles-knowledge is not required for adoption, it can be used to help clarify and create a

context of how to use genetics/genomics in nursing practice.

Items that did not load. The remaining four competency/knowledge items did not load

above .30 on any factor and did not demonstrate Thurstone’s simple structure. The first item

asked if a family history of only first-degree relatives should be collected for all new patients.

This item may not have loaded because while family history information should be collected for

new patients, participants may have been uncertain of how many generations should be included.

The second item that did not load asked about the similarity of DNA between two unrelated

individuals. This item also had a low CVR (- .714), although this may be partly related to

misinterpretation of the question. One content expert stated that the similarity of genetic material

between two individuals was higher than the 90-95%, but that is the content that the question is

addressing. The answer to the question is ‘False,’ because the similarly of genetic material is

closer to 99.5% (Levy et al., 2007). The next item that did not load asked if a single gene variant

causes most common diseases. This item may not have loaded because participants were unsure

of the correct response or the term “variant” in relation to genetic information. The final item

that did not load asked if participants knew about The Essential Competencies and Curricula

Guidelines for Nurses in Genetics and Genomics. This item may not have loaded because it was

not related to a specific clinical topic; while important to address, this item asks about general

awareness of the competencies as opposed to specific genetics/genomics knowledge.

Social systems. Social systems did not load predominantly on a single factor. Of the five

social systems items, two did not load on any factors, two items loaded on multiple factors, and,

as noted, one item loaded on Factor 4 (competency/knowledge 1) (see Table 17). Rogers (2003)

considers diffusion of an innovation as a “kind of social change” (p. 6). New information is

diffused and adopted, leading to changes in the social system. Rogers adds that it may be

difficult to evaluate the role of the social system independently from the characteristics of the

individuals within that system. This illustrates how connected the social system is to the other

variables in the DOI model and parallels the results of the EFA. Social systems may be

pervasive throughout the other variables, as opposed to being an independent variable.

EFA discussion summary. The EFA results indicate that many items did not load on

their GGNPS-designated variables. However, the confidence variable did load as designated in

the GGNPS. In addition, although some item loadings varied, attitude/receptivity,

decision/adoption, and competency/knowledge were clearly identifiable variables in the EFA.

However, there were two distinguishable competency/knowledge factors instead of the

anticipated one. Lastly, social systems items either did not load consistently on an identifiable

factor, loaded on multiple factors, and one item loaded on Factor 4 (competency/knowledge 1).

Content Validity

Eight nursing experts were selected based on current employment, experience, and

evidence of scholarship in nursing and genetics/genomics. Experts with backgrounds in clinical

practice were selected, because the GGNPS is a tool assessing application of genetics/genomics

to clinical practice. Content validity experts rated items in the GGNPS from ‘not relevant’ to

‘very relevant.’ One participant responded as a learner, instead of as a content expert; her

response was excluded. The comprehensive CVI, computed as the mean of the content validity

ratios across all items, was 0.805. This exceeds 0.741, which Wilson and colleagues (2012)

identified as the required threshold when seven content experts evaluate an instrument.

Overall, content experts reported that the GGNPS has content validity. However, they

identified a number of items that were problematic and rated them lowly. Not surprisingly, many

of these problematic items were items that did not perform well on the EFA; loading on two

factors or having low loadings on a single factor. Thus, the results of content validity assessment

and the EFA support each other. The five items discussed below should be considered for

revision or removal.

The lowest CVR rated (- .714) item was a GGNPS-designated competency/knowledge

item P5_Q1: “The DNA sequences of two randomly selected healthy individuals of the same sex

are 90-95% identical”. In addition to having the lowest CVR, this item did not load above .30 on

any factor in the EFA and did not demonstrate Thurstone’s (1947) simple structure.

The next lowest CVR rating (.142) was given to a GGNPS-designated social systems

item P7_Q1_01: “Do you intend to learn more about genetics?” This item loaded on Factor 4

(competency/knowledge 1), with a loading of .308. The item discussed the intention to learn,

which may explain why it loaded on Factor 4 competency/knowledge 1 (how to knowledge).

However, the EFA loading was only slightly above the .30 threshold. One content expert

commented that this item may be misleading because most participants will feel obligated to

answer ‘yes’. To expand on her comment, participants may believe they should learn more about

genetics as health care providers and will answer ‘yes’ out of perceived obligation, as opposed to

intent to learn more about the topic.

Two items had a CVR rating of .428. A GGNPS-designated knowledge-competency item

P6_Q2_02: “In describing your genetic/genomic knowledge, would you consider it to be...”

(participants rated their knowledge as excellent to poor). One content expert noted that this item

may be redundant because the previous item (P6_Q2_01) asks participants to rate their

understanding of genetics of common diseases. The EFA supports this analysis; P6_Q2_01 and

P6_Q2_02 had similar loadings on Factor 4 (competency/knowledge 1) and Factor 1

(confidence). Both items loaded ³. 30 on two factors and had similar loadings for each factor.

Item P6_Q2_01 had a loading of .445 on Factor 1(confidence) and .496 on Factor 4

(competency/knowledge); item P6_Q2_02 had a loading of .445 on Factor 1(confidence) and

.491 on Factor 4 (competency/knowledge 1). Based on these findings, items P6_Q2_01 and

P6_02_02 may be merged into one item focusing on participants’ genetic/genomic knowledge.

The other item with a CVR of .428 was a GGNPS-designated confidence item

P2_Q1_12: “Give patients information about the limitations of genetic testing for common

diseases.” However, this item loaded strongly (.931) on Factor 1 (confidence). One content

expert explained why the item was poorly rated in the CVR, stating providing this information

100

may be outside of the scope of practice of nursing. Instead, nurses would reinforce information

provided by genetics specialists and provide clarification.

One additional item, included in content analysis, was not a part of the GGNPS used in

this study. Rather it was added when the GGNPS was edited after pilot testing. This item also

had a low CVR rating of .428. This item: “Thinking about how you support clinical decisions

(such as administering drugs prescribed), how important do you think each of the following is to

consider? Genetic Test Result,” while asking about very important content, is complex. Some

content reviewers were unsure if the item was asking about medication administration

specifically or medication administration was just one example of a clinical decision. Genetic

testing results may be considered when administering a number of medications, such as codeine

and warfarin (Kitzmiller, Groen, Phelps, & Sadee, 2011). However, the question seems to be

asking about general clinical decision-making. One content reviewer provided helpful insight,

stating an example may clarify this item.

In summary, while many items had very high CVR’s, the content experts identified a

number of items that may be unclear or may be misinterpreted by individuals completing the

GGNPS. Content validity results also shed light on why construct validity could not be

established. Several items that were scored lowly by content experts also performed poorly on

the EFA. Items with low CVRs that also loaded on multiple factors in the EFA could have

affected the results of a CFA because the item(s) likely attempted to load on multiple factors,

increasing the likelihood that parameter estimates could not be provided by MPLUS. Items that

scored lowly and did not load on any factor in the EFA could affect the SEM and CFA; making it

more difficult to identify the factors in the model and the relations between them. Content expert

analysis also provides direction for future assessments of construct validity. Items that

performed poorly can be considered for revision or even removal from the instrument.

101

Following this evaluation, the CFA and SEM can be reassessed.

Face Validity

Face validity was conducted, with seven reviewers, to evaluate the ease of understanding

and relevance of the GGNPS to nursing practice. All reviewers were currently practicing nurses

with a variety of educational and clinical backgrounds. None of the reviewers had specific

education or experience in genetics/genomics; they were purposely selected to provide the

perspective of the majority of nurses in the workforce.

Five of the seven reviewers reported the GGNPS was somewhat clear/easy to understand;

the remaining two reviewers reported that it was very clear/easy to understand (see Table 7).

The majority of reviewers (n = 5/7) also reported the GGNPS items were pertinent to their

clinical practice, while two reviewers reported that they were not relevant. However, among the

two face validity reviewers who reported that the GGNPS was not pertinent to their clinical

practice, one reported that it was very clear/easy to understand. There was not a clear pattern

among reviewers of understanding the content of the GGNPS and realizing that

genetics/genomics applies to nurses across all specialties and practice settings. This indicates a

knowledge gap regarding the use of genetics/genomics in clinical practice. In fact, four out of

seven face validity reviewers reported that they did not use genetics/genomics in their clinical

practice, while three out seven reported that they did (see Table 7).

Overall feedback from the reviewers indicated that the GGNPS has face validity. None of

the participants reported that the GGNPS was ‘not very clear’ or ‘unclear and difficult to

understand.’ In addition, five out of seven nurses from a variety of clinical specialties, roles,

educational backgrounds, and levels of experience reported that the GGNPS was pertinent to

their clinical practice. One of the goals of this instrument is to make it applicable for a variety of

nurses, regardless of specialty, clinical role, or level of education (Calzone et al., 2014).

102

Construct Validity

As noted, construct validity could not be established because the CFA failed to converge.

Furthermore, the SEM showed that the hypothesized model did not fit the data and many

relations in the SEM were weak in light of the large sample size. Please note, standardized units

of measurement for the SEM are discussed in Appendix G; this information is presented in the

appendix because β weights for nominal and ordinal data are complex and require an extended

discussion.

Confirmatory factor analysis (CFA). A CFA, using the original data and imputed data,

was conducted to evaluate whether the GGNPS items comprised the underlying variables of

confidence, competency/knowledge, attitudes/receptivity, decision/adoption, and social systems.

After over 34 hours and 460 iterations, MPLUS reported that the models did not converge on a

stable set of parameter estimates. This occurred because the hypothesized GGNPS variables

mentioned above did not fit the data. The items from the GGNPS that were thought to make up

the variables of confidence, competency/knowledge, attitudes/receptivity, decision/adoption and

social systems could not be grouped this way despite multiple attempts by the statistical

software. As discussed, EFA provided further evidence of this because items that were thought to

load on one factor, frequently loaded on another factor.

As a result, no factor loading can be discussed. This highlighted the need to conduct an

EFA in the ancillary analysis. A five-factor model was selected to parallel Rogers’ (2003) DOI

model and the EFA produced a data-driven analysis of how each item from the GGNPS loaded

on the factors. EFA results indicated that GGNPS-designated confidence items loaded on one

factor; however, GGNPS-designated competency/knowledge, attitudes/receptivity, and

decision/adoption items were more interrelated than originally thought. While

competency/knowledge, attitudes/receptivity, and decision/adoption were identified as variables

following the EFA analysis, some items thought to be associated with one variable actually

103

loaded on a different variable. By contrast, social system was not an identifiable variable;

GGNPS-designated social systems items either did not load on any factors, loaded on multiple

factors, and one item loaded on competency/knowledge.

Structural equation model (SEM). A SEM was used to explore overall model fit and

the relations among the variables in the data. Evaluation of a number of overall fit parameters

(Chi-square, RMSEA, CFI, and BIC) all indicated that the study’s hypothesized model did not fit

the data. The poor fit was consistent in both the original data set and all four imputations.

Further supporting the poor fit is the significant amount of variance (residual) that was not

explained (see Table 10). The unexplained variance may be due to relations among variables that

were not considered in the hypothesized model. The poor fit of the data to the hypothesized

model, based on work by Calzone and colleagues (2012), combined with the large amount of

unexplained variance, again illustrated the need to conduct an EFA in ancillary analysis. The

EFA can help increase understanding of both the variables themselves and can be used to

examine proposed relations between the variables.

Figure 9

p* < 0.001; p** = 0.013

SEM Using the Original Data Set.

0.104*

Social Systems

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

.391*

0.150*

0.717*

0.399

Confidence

0.039**

0.225*

104

Not only did the data not fit the hypothesized model, many of the relations among the

variables in the SEM were weak. Because a large sample size increases the likelihood that

statistically significant correlations exist among the data, the SEM results must be carefully

evaluated (Wolf, Harrington, Clark, & Miller, 2013). The weak relations may be partly due to

conceptual issues. That is, the hypothesized model above was based on Rogers’ DOI model, but

it was revised for this study to include confidence as a variable and exclude components of the

DOI that were not measured in the GGNPS. The weak relations may also be explained in part by

the EFA. The EFA showed that many items designated in the GGNPS as one variable, actually

loaded on a different variable. Even if a small number of items in each variable, other than

confidence, actually measured a different variable this would create measurement error. This

measurement error would then be compounded because the SEM evaluated direct and indirect

relations amongst the five variables.

Competency/knowledge on attitudes/receptivity. An example of a statistically

significant, but weak, relation is the direct effect of competency/knowledge on

attitudes/receptivity (.104, p < 0.001) (see Figure 10). This is a surprising finding because

according to Carroll and colleagues (2009) increasing knowledge has a positive effect on

attitudes/receptivity. Researchers conducted an educational intervention that positively affected

the attitudes about genetic testing of participants including nurses, nurse practitioners,

physicians, and other health care providers. Following an educational workshop and modules,

researchers reported a statistically significant increase in the number of participants agreeing that

genetic testing was beneficial in the management of adult-onset diseases (p= .031), using the

McNemar test for matched categorical variables (Carroll et al., 2009).

A number of factors may explain this surprising finding. According to Rogers (2003),

while knowledge is predominantly 'knowing' or cognitive, attitude/receptivity is predominantly

105

'feeling'. Rogers adds that attitude is “a relatively enduring organization of an individual’s

beliefs about an object that predisposes his or her actions” p. 174-175. For example, if an

individual associates use of genetics/genomics in their clinical role with stress or uncertainty,

increased knowledge will not fully overturn this enduring attitude. In fact, results show that the

hypothesized model was not correct, increased knowledge has little effect on one’s attitude.

The weak relation may be partially attributed to the EFA results as well. A number of

GGNPS-designated attitudes/receptivity items loaded on decision/adoption; these items may

have created measurement error. Items that were actually more related to the decision/adoption

variable were evaluated in the SEM as part of the attitudes/receptivity variable, impacting the

model’s ability to show accurate relations between variables.

Results of reliability testing should also be considered in assessing relations between the

variables. A number of attitude/receptivity items, especially items that were “check all that

apply” or included large Likert scales were the most poorly performing items during reliability

evaluation (Calzone et al., 2016). The low reliability of these items led researchers to revise or

eliminate a number of those questions. Therefore, relations that include attitude/receptivity items

should be evaluated with the knowledge that a number of attitude/receptivity items performed

poorly during reliability testing and were amended or eliminated as a result.

Competency/knowledge on confidence. The direct effect of competency/knowledge on

confidence in using genetics/genomics in one’s practice was a stronger (.399, p < 0.001) relation,

meaning increased knowledge about genetics/genomics increased confidence (see Figure 10).

This relation is supported by Benner (1984), who noted that experienced (knowledgeable) nurses

have more confidence in their abilities and are more capable of managing difficulties and

challenges. As nurses gain knowledge and increase their mastery of content, they increase their

critical thinking ability. Critical thinkers, therefore, exhibit a number of characteristics including

106

confidence (Benner et al., 2008). Thus, it is plausible that nurses who have more knowledge of

genetics/genomics are more confident in the use of this knowledge.

This relation was also supported empirically; Silva and colleagues (2016) evaluated the

effect of knowledge on confidence. Study participants completed an educational program about

suicide risk assessment and management. Participants consisted of 2,843 nurses, physicians,

case managers, administrators, and support staff. Those who completed the educational training

had statistically significantly higher confidence in their ability to assess patients and provide

interventions than those who did not receive the training (F [1,1333] = 298.21). This supports

the results of the hypothesized SEM model, which found a substantial relation between

competency/knowledge and confidence.

As previously noted, items in the GGNPS that Calzone and colleagues (2012) designated

as part of the confidence variable, all loaded strongly on the confidence factor. Thus, it is likely

that these items accurately measure confidence. Potential measurement error may have resulted

from GGNPS-designated competency/knowledge items that loaded on Factor 2

(attitudes/receptivity) in the EFA. According to the EFA results, those items were more related to

the attitudes/receptivity variable as opposed to the competency/knowledge variable; as a result,

they may have impacted the strength of this relation.

Competency/knowledge on decision/adoption. The indirect effect of

competency/knowledge on decision/adoption (-.023), through attitudes/receptivity, (not shown in

diagram see Table 11) was negative, weak. As described by Rogers (2003), diffusion of

innovation is a process whereby an individual initially acquires knowledge about an innovation

and forms a favorable or unfavorable attitude about the innovation. This is the direct effect of

knowledge on attitudes/receptivity discussed above. After attitudes are formed as a result of this

knowledge, the person decides to adopt or reject it.

107

However, in a clinical setting/context, additional factors may influence adoption, such as

limited time/resources and competing responsibilities. A previously described study supports this

point. Cook and colleagues (2013) evaluated the effect of an educational intervention on the

knowledge, attitudes, confidence, and clinical practice of nurses providing care for patients with

mild traumatic brain injury (mTBI). Following the educational intervention about new clinical

guidelines, trauma nurses reported being more able to identify interventions to alleviate

symptoms following an mTBI (χ² = 6.125, p = .001)(Cook et al., 2013). Upon assessing nurses’

attitudes, 96% (n = 24/25) of participants reported the educational intervention (knowledge)

increased their understanding of symptom assessment and new clinical guidelines. However,

while between-group analysis showed that some nurses increased their use of mTBI symptom

assessment (50-86%), only 52% of the participants reported using mTBI symptom assessment in

their clinical practice (adoption) (Cook et al., 2013). Additional factors, such as social systems,

that relate to nurses’ responsibilities and work stressors may affect the indirect relation of

competency/knowledge and decision/adoption, through attitudes/receptivity.

Finally, the EFA results indicated that competency/knowledge loaded on three factors and

three GGNPS-designated decision/adoption items loaded on Factor 4

(competency/knowledge 1), this measurement error may partially account for the weak relation

between the two variables.

Attitudes/receptivity on decision/adoption. In fact, the direct negative effect of

attitudes/receptivity on decision/adoption was stronger and statistically significant (-.225, p <

0.001) (see Figure 10). Although this is a surprising finding, Rogers (2003) states that one’s

attitude or receptivity to an innovation is enduring. GGNPS-designated attitude/receptivity items

ask nurses about the perceived advantages and disadvantages of genetics/genomics. This effect

108

may indicate that perceived disadvantages have a substantial negative effect on the decision to

adopt genetics/genomics into nursing practice.

Examples of disadvantages evaluated in the GGNPS include genetics/genomics taking too

much time and being an additional burden for nurses. Nursing literature often sites a lack of time

and overwhelming burden of responsibilities as a challenge for nurses in many clinical practice

settings (Sussner et al., 2011; Tourangeau et al., 2013). The perceived disadvantages of

genetics/genomics can lead to negative attitudes/receptivity toward the innovation and have a

negative effect on its adoption. To promote adoption of genetics/genomics, clinical agencies

must promote positive attitudes about this innovation; however, this must be done in context of

health care institutions providing support for nurses. This support may include addressing

workplace stressors and creating support networks to promote education and best-practice

initiatives. This also demonstrates the pervasiveness of the social system variable. The role of

health care organizations is further discussed below when evaluating the relation between

attitudes/receptivity and social systems.

However, as previously mentioned, a number of attitude/receptivity items performed

poorly during reliability evaluation and were amended or eliminated as a result. Therefore,

relations that include attitude/receptivity items should be evaluated with the knowledge that both

the poor performance of these items during reliability testing and their revision could affect the

relation between attitudes/receptivity and decision/adoption.

This relation must be cautiously interpreted, eight out of 17 GGNPS-designated

attitudes/receptivity items loaded on Factor 3 (decision/adoption) (see Table 14). Since so many

items that were thought to represent the attitudes/receptivity variable actually loaded on

decision/adoption, this increases the risk of measurement error and the relation between these

two variables should be reevaluated after reclassifying the variables based on the EFA results.

109

Social systems on competency/knowledge. The direct effect of social systems on

competency/knowledge (.717, p < 0.001) was statistically significant and strong (see Figure 10).

This supports Rogers (2003) assumption that diffusion of innovations is a process that is

encompassed within a social system; diffusion of innovations is “communicated through certain

channels over time among members of a social system” (p. 5). He also summarizes the diffusion

of innovations process as essentially an “information-seeking and information-processing

activity” (Rogers, 2003, p. 172). In fact, knowledge is the first step of the diffusion of

innovations process and has indirect effects on all of the following steps. As a result, both social

systems and knowledge are strongly integrated throughout the DOI process.

An example of the effect of social systems on competency/knowledge is demonstrated in

the nursing faculty champion initiative study (Jenkins & Calzone, 2014). Acquisition of

genetics/genomics knowledge required not only faculty expertise, but support from university

administrators and peer networks. University deans and program directors championed faculty

applicants to participate in the initiative and peer support networks provided an additional source

of social system support by using faculty genetics/genomics experts as resources for learners.

Results of the initiative indicate that the faculty champion model was effective in promoting

integration of genetics/genomics in nursing curricula; however, researchers did note a need for

additional communication with participants to maintain engagement. Initiatives such as the

faculty champion program can be important mechanisms to help faculty reach the educational

genetics/genomics-related goals outlined by the ACCN and CCNE.

The EFA results provide a caveat when interpreting the relation between social systems

and competency/knowledge. Three out of five GGNPS-designated social system items loaded on

other factors and two did not load on any factor. In addition, GGNPS-designated

competency/knowledge items loaded on three factors in the EFA. The strength of the relation

110

reported here does not include additional insights provided by the EFA. In fact, all of the

relations including social systems require further exploration because the majority of

GGNPdesignated social system items loaded on different factors. Further studies are required to

amend the GGNPS and reevaluate the relations among the variables.

Social systems on attitudes/receptivity. In contrast, the direct effect of social systems on

attitudes/receptivity (.150, p < 0.001) was much weaker, although statistically significant (see

Figure 10). Indicating that support from the social system has a smaller than anticipated role in

promoting a positive attitude towards genetics/genomics. This is an unexpected finding since

Rogers (2003) describes both the pervasive nature of social systems and how one’s attitudes

toward an innovation are relatively enduring. This relation requires further exploration in future

research, because in the EFA, a number of attitude/receptivity items loaded on factors other than

the GGNPS-designated attitude/receptivity variable. In addition, GGNPS-designated social

systems items loaded on a variety of factors. This relation also requires further exploration

because a number of attitude/receptivity items performed poorly during reliability testing and

were amended or eliminated as a result.

The relationship between social systems and attitudes/receptivity may be weak due to the

social climate in the workplace and how it affects nurses’ attitudes/receptivity to innovations.

Nurses often receive mixed messages, on one hand current and patient-centered clinical practice

are emphasized, but in reality nurses are pressured to prioritize responsibilities emphasized by

the clinical institutions. For example, medication administration within a specific time frame.

Tourangeau and colleagues (2013) described inadequate time and poor staffing as frequent

nursing stressors. Not only are nurses are required to administer medications and answer call

bells in a timely manner, but they are often caring for a high patient load. In addition,

benchmarks such as patient satisfaction are continuously monitored by instutions. Health care

111

organizations prioritize measures such as patient satisfaction, evaluated in the HCAHPS survey,

partially because it determines reimbursement of over $850 million dollars annually (Zusman,

2012). The reality of nursing is that nurses are often faced with a high patient load and numerous

time pressures, while still being encouraged to maintain a current clinical practice that reflects

innovations such as genetics/genomics. While many nurses may strive to maintain a clinically

current practice, stressors from the social system of the health care organization are likely to

cause ambivalnce regarding complex innovations such as genetics and genomics. This is further

emphasized by results of the MINC study, in which 25.3% (n = 1,342/5,314) of participants

reported they believed senior staff members considered genetics an important part of the

participants’ role. Genetics/genomics may compete with other responsibilities and do not

directly affect reimbursement for RNs, thus resulting in the weak relation between social systems

and attitudes/receptivity.

Social systems on decision/adoption. While the direct effect of social systems on

decision/adoption was relatively strong (.391, p < 0.001) (see Figure 10), the indirect relation

between social systems and decision/adoption (-.034) was a negative and weak relation (see

Table 11). That is, social system support does promote adoption of genetics/genomics by nurses.

However, social system support, as an indirect relation through attitudes/receptivity, does not

contribute to adoption of genetics/genomics. Rogers (2003) describes diffusion of innovations as

a process encompassed within the social system. He also adds that norms of the social system

establish patterns of behavior among its members. Therefore, the relation between social

systems and decision/adoption is expected to be strong. The indirect relation through

attitudes/receptivity is expected to be weak because the relation between social systems and

attitudes/receptivity was also weak.

112

This is seen clinically in a study by Shields and colleagues (2010). The authors

highlighted that social system factors may prevent correct diagnosis of maturity onset diabetes of

the young (MODY). Geographical areas with greater research interest in MODY have higher

rates of diagnosis and referral. Social systems in those health care organizations may provide

additional support because research interests bolster vigilant clinical screening for conditions

such as MODY. The weakness of the indirect relation observed in this study may be due to the

moderating effect of attitudes/receptivity. Nurses' attitudes/receptivity are often related to the

perception of the innovation as related to competing work demands and stressors. For example,

nurses report inadequate time as a major barrier to collecting family history information, further

demonstrating that high workload and inadequate time are perceived disadvantages of

genetics/genomics (Sussner et al., 2011). A study using Rogers' DOI, although unrelated to

genetics/genomics, also identified time pressures perceived by nurses (Lee, 2004): there, the DOI

model was used to evaluate nurses’ perceptions of the innovation of computerized care systems.

Time shortage as a stressor was identified as a key theme. This stressor negatively affected

nurses’ attitudes toward the innovation.

However, similarly to other relations including social systems, it is important to consider

that EFA results indicated the GGNPS-designated social system items loaded on a variety of

other factors and some GGNPS-designated decision/adoption items loaded on

competency/knowledge. The relations must be reevaluated in further studies.

Confidence on attitudes/receptivity. The direct effect of confidence on

attitudes/receptivity (.039, p = 0.013) was the weakest relation in the SEM and was not

statistically significant in imputed data set 1 (see Figure 10 and Appendices F.1-F.4). This

indicates that confidence does little to create a positive attitude or increase receptivity towards

genetics/genomics; it is an anticipated finding because confidence has a weak effect on creating a

113

positive attitude towards genetics/genomics. This could be related to other variables affecting the

relation between confidence and attitudes/receptivity. For example, Metcalfe and colleagues

(2009) evaluated the confidence and attitudes of hospice nurses in their use of genetics in nursing

practice. Initially the attitude of 43% (n = 43/100) of nurses was that adopting genetics into

hospice care was of ‘little importance’; only 3% (n = 3/100) were confident enough to teach

clients. However, after reading scenarios about applications of genetics in hospice care, 66- 99%

rated genetics as ‘very important’ or ‘essential’ (Metcalfe et al., 2009). A key consideration in

this study is that knowledge may have also affected nurses’ attitudes; the intervention that

increased nurses’ attitudes/receptivity included education about genetic/genomic applications.

This provides insight into the weak relation between confidence and attitudes/receptivity seen in

this study; the weak relation between confidence and attitudes/receptivity may also be affected

by additional variables, such as competency/knowledge.

The EFA results also help explain the weak relation between confidence and

attitudes/receptivity. Although GGNPS-designated confidence items loaded consistently on

Factor 1, eight out of 17 GGNPS-designated attitudes/receptivity items actually loaded on Factor 3

(decision/adoption). As a result, the SEM may not accurately evaluate the relation between

confidence and attitudes/receptivity.

In addition, as previously mentioned, a number of attitude/receptivity items performed

poorly during reliability evaluation and were amended or eliminated as a result. Therefore,

relations that include attitude/receptivity items should be evaluated knowing that both the poor

performance of these items during reliability testing and their revision could affect the relation

between attitudes/receptivity and confidence.

Confidence on decision/adoption. The indirect effect of confidence on decision/adoption

(.0008) is the weakest indirect relation in the SEM (see Table 11). That is, confidence, through

114

attitudes/receptivity, has very little effect on the decision to adopt genetics/genomics into nursing

practice. Confidence is not an explicit part of the DOI model; however, Rogers (2003) does

address how personality aspects can affect adoption of an innovation. He describes earlier

adopters as better able to cope with uncertainty and risk. Benner’s (1984) research regarding

confidence paralleled Rogers’ views; she noted experienced nurses have more confidence in their

abilities and are better capable of managing difficulties/challenges.

However, this is a surprising finding because multiple studies have found a relationship

between confidence and decision/adoption. Barr and McConkey (2006) noted that low

confidence affected adoption of genetics/genomics by nurses and midwives. The study defined

adoption as providing referrals and patient education. Only 2.5% (n = 5/194) strongly agreed

and 10.8% (n = 21/194) agreed that they were comfortable referring clients to a genetic

counselor; 1.5% (n = 3/194) strongly agreed and 7.2% (n = 14/194) agreed that they were

comfortable discussing test results of pediatric patients. This was supported by findings by

Bottorff and colleagues (2005) who reported that decreased confidence was associated with

decreased adoption of genetics/genomics by nurses (n = 975); researchers defined adoption as

providing referrals, gathering genetic information, and obtaining informed consent for releasing

genetic testing information.

This surprising finding may be partially explained by the fact that the indirect relation

between confidence and decision/adoption could be affected by attitudes/receptivity. Negative or

mixed attitudes/receptivity towards genetics/genomics could affect adoption even if individuals

increase their confidence. In addition, a number of items originally designated as

attitudes/receptivity and decision/adoption in the GGNPS did not load on these variables in the

EFA. The relation between confidence and decision/adoption should be further explored in

115

future research that will reevaluate a CFA and SEM, following modifications based on EFA

results.

EFA results similarly help to explain the weak relation between confidence and

decision/adoption. The GGNPS-designated confidence items loaded consistently on Factor 1,

but three out of 12 GGNPS-designated decision/adoption items loaded onto Factor 4

(competency/knowledge 1) (see Table 15). In addition, the indirect effect of confidence on

decision/adoption is through attitudes/receptivity. Eight out of 17 GGNPS-designated

attitudes/receptivity items actually loaded on Factor 3 (decision/adoption). The measurement

error in each variable may be compounded as the relation between variables is evaluated.

Additional Considerations

There are two additional considerations in evaluating the SEM results. First, as

previously mentioned, the test-retest reliability analysis conducted by Calzone and colleagues

(2016) resulted in edits to several attitude/receptivity questions, because they were some of the

worst performing items in the reliability analysis. The SEM results of this study also demonstrate

that attitude/receptivity items contributed to the weakest relationships in this analysis. Evaluation

of Figure 9 demonstrates that variables endogenous and exogenous to attitudes/receptivity were the

weakest relationships in the SEM. Examples include the relationship between

competency/knowledge and attitudes receptivity (.104); attitudes/receptivity and decision/adoption

(-.225); confidence and attitudes/receptivity (-.039) and social systems and attitude receptivity

(.150). Relations that included attitudes/receptivity as either the dependent or independent variable

were all weaker than relations that looked at variables other than attitudes/receptivity.

By contrast, the SEM results support the strength of the effect of social systems.

Although the results of the EFA demonstrate that social system items did not load as a cohesive

variable, evidenced by most social system items either not loading above .30 on a variable or

116

loading on multiple variables, social system items account for two strong relationships within the

SEM. Specifically, the relationships of social systems and competency/knowledge (.717) and

social systems and decision/adoption (.391). In the SEM analysis, each variable was evaluated

using a factor score based on the GGNPS-designation. For example, the social system factor

score was determined by combining the participants’ responses to the GGNPS-designated social

system items. GGNPS-designation was used for all variables in the SEM because it was based

on previous research by Calzone and colleagues (2012, 2014), based on their analysis of Rogers’

DOI theory and evaluation of the Essential Competencies and Curricula Guidelines for Nurses in

Genetics and Genomics. The strength of relations in the SEM that include social systems,

compared to the weak performance of the social system items in the EFA, indicate that although

social systems may not be a standalone variable, social system items do explain a significant

portion of variance in the analysis and play an important role in understanding the relationships

between the variables. In further studies, social systems may be explored as a moderating factor

or social system items can be evaluated individually, as opposed to as a cohesive variable.

Summary Analyses support that the GGNPS has both content and face validity. However,

construct validity remains to be established. This study’s proposed CFA and SEM models did

not fit the data. EFA results, in ancillary analysis, shed light on likely sources of measurement

error and provide the basis for recommendations concerning revision and retesting of the

GGNPS.

117

Chapter VI

Conclusions, Implications, and Recommendations

Findings from this study met the threshold for both content and face validity for the

GGNPS. Although construct validity could not be established, the findings provide insights that

can guide future evaluation of construct validity. This chapter discusses conclusions and

implications for nursing practice, research, and education. It also addresses limitations of the

study and goals of future research.

Conclusions

Content validity. The comprehensive CVI, computed as the mean of the content validity

ratios across all items, met the threshold required to establish content validity. However, the

content experts identified a number of items in the GGNPS that may require revision. In fact,

many of these same items did not perform well in the EFA. Information from the content

validity analysis and the corroborating EFA results can be used to reevaluate those items.

Furthermore, this information can be used in future analyses of construct validity of the GGNPS.

Face validity. Findings from this study also support that the GGNPS has face validity.

Face validity was established by evaluation of the GGNPS for ease of understanding and

relevance to nursing practice.

118

Construct validity. The results of this study cannot establish construct validity based on

the CFA and SEM results. The hypothesized model could not converge on a stable set of

parameter estimates and the CFA did not produce factor loadings. This occurred because the

GGNPS-designated confidence, competency/knowledge, attitudes/receptivity, decision/adoption,

and social systems items did not load on those corresponding variables. The EFA provided

further evidence of this because many items that were designated as one factor loaded on another

factor. Additionally, the SEM results indicated that the hypothesized DOI model did not fit the

data, and many relations in the SEM were weak in light of the large sample size.

Ancillary analysis. The EFA provided significant insight into this study’s main analyses.

While confidence items loaded as predicted in the EFA, multiple items in the

attitudes/receptivity, competency/knowledge, and decision/adoption variables unexpectedly

loaded on other variables. In addition, consistent with Rogers’ DOI theory of different types of

learning related to adoption of innovations, competency/knowledge loaded on two factors.

Lastly, social systems did not load on a unique factor. Instead, some items loaded on multiple

factors, some did not load on any factors, and one item loaded on competency/knowledge.

The implications of the EFA findings are further discussed below as implications for future

research on the GGNPS.

Limitations

Several limitations must be noted when considering this study’s failure to establish

construct validity including: hypothesis of an incomplete SEM model, subject burden, and data

missingness.

Incomplete SEM model. The use of a SEM to establish construct validity was limited

because the GGNPS does not purport to measure several components of Rogers’ DOI model.

Specifically, the GGNPS does not measure prior conditions, other than social systems, that may

119

affect adoption of new innovations. According to Rogers (2003), prior conditions also include

previous practice, felt needs/problems, and innovativeness.

The GGNPS also does not measure characteristics of a decision-making unit, as described

by Rogers (2003), and their potential effect on adoption of genetics/genomics. These

characteristics include socioeconomic characteristics, personality variables, and communication

behavior. Socioeconomic characteristics related to adoption of genetics/genomics may include

the ability to fund nursing education (e.g. continuing on to a baccalaureate or master’s degree)

and/or means to purchase continuing education courses. Rogers (2003) reports that personality

variables are studied less often partly because it is difficult to measure aspects of someone's

personality. However, these variables can affect the rate of adopting an innovation. Rogers gives

examples of the personality traits that promote early adoption include being open to new ideas,

having greater ability to think abstractly, having a favorable attitude towards change, being able

to cope with uncertainty and risk, and having a positive attitude towards scientific research.

Based on the DOI model, communication behavior also affects adoption of innovations.

According to Rogers, early adopters are more connected to members of their social systems, seek

information more actively, and have greater knowledge regarding innovations.

At present, the GGNPS does not include items that evaluate the characteristics of a

decision-making unit or prior conditions. The GGNPS contains 72 items and demographic

questions; therefore, adding items may make the instrument too long and cumbersome to

complete. Instead, as discussed below in future analyses, additional insights into characteristics

of decision-making units and prior conditions can be gathered via focused studies. For example,

qualitative analyses highlighting these portions of Rogers’ theory.

However, it is important to mention that the GGNPS was shortened following reliability

testing. Poorly performing items were reformatted or, if other, better performing items, measured

120

the same domain, the poorly performing item was eliminated (Calzone et al., 2016). Items from

three “check all that apply” questions were eliminated, those items were part of the

attitudes/receptivity domain. Four confidence items were eliminated. A component of one

decision/adoption item was eliminated. A competency/knowledge item discussing the genetic

risk of lung cancer was removed. Lastly, two questions, one addressing the preferred format for

continuing education and one about how family history is collected were deleted and therefore

not mapped to domains. Therefore, if focused studies about decision-making units or prior

conditions yielded significant insights, that content could be considered for addition to the

GGNPS.

Subject burden. Survey burden and time constraints are common, but important,

limitations in clinical research. RNs may have completed the GGNPS during a busy shift and

had inadequate time to consider all of the items carefully. Some may intend to participate but

realize they do not have adequate time, others may feel pressured by management to participate

and do not answer to the best of their ability. Of the total 7,798 participants, 679 (8.71%)

respondents opened the instrument to complete it but left all items blank. In addition, 100% of

the institutions used for data collection were concerned with survey burden and had conducted

institution-wide nursing data collection in the past six months (Calzone et al., 2014). Thus, the

accuracy of responses may have added to measurement error.

Missing data. The data set contained a substantial amount of missing responses.

Welch’s two-sample independent t-tests were used to determine if data could be considered

missing at random (MAR). Missingness on decision/adoption items could not be considered to

be MAR; however, missingness on other variables did appear to be MAR. As a result, because

all data are not MAR, multiple imputation was used to account for missing items. Multiple

imputation replaces missing data with values imputed from the values of all existing data.

121

Imputation is conducted for all items; the resulting imputed data sets are reevaluated and

compared to the original data set and each other.

Multiple imputation is the most statistically robust method to account for data not missing

at random (Carpenter et al., 2006). Therefore, although there was a significant amount of data

missing in the participants’ responses, the missing data was managed via the most statistically

rigorous method. Although multiple imputation is a statistically robust method, imputed data is

extrapolated. It is modeled statistically to approximate participant responses as much as

possible; however, the statistical inferences may increase the likelihood of measurement error.

This is a limitation in the study because there is a possibility that the results using imputed data

could differ from how participants would respond if they completed the instrument in its entirety.

Implications

GGNPS revisions. The results of this study provide a number of insights that can be

used to guide further revisions of the GGNPS. Several items that were rated lowly by content

experts also performed poorly on the EFA. These items may be considered for revision or

removal.

Item P5_Q1 (“The DNA sequences of two randomly selected healthy individuals of the

same sex are 90-95% identical”) was a ‘True/False’ question, the correct response was ‘False.’

This item was the lowest rated item in the CVI (CVR = -.714). However, the low CVR may be

related to confusion among the experts. For example, one content expert indicated that the

percent similarity of DNA sequences is higher than 90-95%. The correct answer is ‘False’

because genetic similarity is closer to 99.5% (Levy et al., 2007). Five out of seven content

experts rated the item and ‘not relevant’ or ‘somewhat relevant,’ only two rated it as ‘relevant’ or

‘very relevant.’ However, this feedback should be cautiously considered, because other content

experts may have misread the question as well. Finally, this item did not load above .30 on any

122

factor in the EFA and did not demonstrate Thurstone’s (1947) simple structure. Despite the fact

that this item performed poorly on the EFA, recommendations to revise or remove this item

cannot be made without additional discussion with content experts due to possible

misinterpretation of the item.

Item P7_Q1_01 (“Do you intend to learn more about genetics?”) had one of the lowest

CVRs (.412). Three out of seven content experts rated the item and ‘not relevant’ or ‘somewhat

relevant,’ the remaining four rated it as ‘relevant’ or ‘very relevant.’ One content expert stated

the item may be misleading because all participants are likely to answer ‘yes.’ The item loaded

.308 on Factor 4 (competency/knowledge) and demonstrated Thurstone’s simple structure.

However, among the 7,798 responders, n = 2,540 replied ‘no’ and n = 2,751 replied ‘yes.’ The

remaining, n = 2,507, left the item blank. Therefore, participants were not more likely to answer

‘yes’ and did not seem to feel compelled to state they intend to learn more about genetics.

Question 2 in Part 7 of the instrument asked a follow up question about how helpful participants

found the following resources (e.g. books, web-based information, lectures, workshops) for

learning about genetics. However, this question was removed when the instrument was revised.

Furthermore, this question does not address accessibility. For example, a nurse may perceive

inperson workshops as very helpful, but only have time and financial resources available for

seeking out web-based information). Authors of the GGNPS may consider a follow up question

to provide insight into resources participants are most likely to use. This information can also be

used to create resources (e.g. webinars, in-person courses, clinical cases distributed via email

weekly, etc.) based on learner demand.

Table 19

123

Items with Low Factor Loadings and Associated CVRs

Item

Highest

Simple

Structure

CVR

P1_Q1

“How important is it for the nurse to become more

educated about the genetics of common diseases?”

.282

Absent

P1_Q3_04

“Potential disadvantages: would not improve patient

care.”

.292

Absent

N/A

P2_Q3_01

“A family history that includes only 1st degree relatives

such as… should be taken for every new patient.”

(shortened)

-.173

Absent

.714

P5_Q1

“The DNA sequences of two randomly selected healthy

individuals of the same sex are 90-95% identical.”

.118

Absent

-.714

P5_Q2

“Most common diseases such diabetes and heart

disease are caused by a single gene variant”

.132

Absent

P6_Q1

“Have you heard of the Essential Competencies… in

Genetics and Genomics?” (shortened)

.281

Absent

0.714

P7_Q1_02

“Would you be able to attend a course during work

hours?”

.116

Absent

P7_Q1_03

“Would you be able to attend a course on your own

time?”

.215

Absent

.714

A number of items did not load above .30 on any factor and did not demonstrate

Thurstone’s simple structure (see Table 19). However, most of those items, with the exception of

P5_Q1, were highly rated by the content experts. Additionally, the content experts did not

review item P1_Q3_04 because it was reworded when the GGNPS was initially revised. The

items that did not load above .30 on any factor may be considered for revision; however, these

results may also highlight a knowledge gap in particular content areas. An EFA evaluates shared

variance among data, if many of the participants answer an item incorrectly, with the same

124

incorrect answer, this decreases the covariance of this item with other items. As a result,

decreasing the ability of that item to load above .30 on any factor. Therefore, items must be

analyzed using the EFA results, the content expert analysis input, and evaluation of the topic

covered in the item.

For example, P2_Q3_01 (“A family history that includes only 1st degree relatives such as

parents, siblings, and children should be taken for every new patient”) addresses family history

collection, one of the most important components of evaluating genetic information. Not only

does this item address a key genetic/genomic concept, it also has a high CVR (.714). This item

may have loaded poorly (-.173) because participants did not know 1) how many generations are

included in a family history and/or 2) how frequently family history information should be

collected. The poor loading may be related to a knowledge deficit in one or both of the topics

addressed in the item. Additionally, it may be related to the question compounding two

concepts, if participants do not know the answer to either concept it could affect their response.

Item P5_Q2 (“Most common diseases such diabetes and heart disease are caused by a single

gene variant”) had a highest loading of .132. However, it had an extremely high CVR of 1. This

item should not be removed from the GGNPS because it addresses information that is directly

applicable to nursing practice. The low loading, but high CVR, may be related to a lack of

knowledge regarding the topic. The poor loading may also be related to the wording of the item;

participants may not understand the term ‘gene variant’ as opposed to ‘gene.’ Lastly, items

P7_Q1_02 (“Would you be able to attend a course during work hours?”) and P7_Q1_03

(“Would you be able to attend a course on your own time?”) did not load above .30 on any

factor and did not demonstrate Thurstone’s simple structure. However, they had high CVR

scores (see Table 19). While these items gather important background information about

availability and accessibility of education in genetics/genomics, these items are not clearly

125

associated with social systems, even though they are designated as social system items in the

GGNPS. For example, while attending a course during work could be related to the support a

health care setting (social system) provides, busy clinical units may not be able to accommodate

education during clinical time due to the demands of patient care. As a result, this item may not

accurately evaluate social system support. Similarly, attending a course outside of work may be

related to other time pressures such as family and/or personal responsibilities. Therefore, these

items provide very relevant information, but may be considered as part of

demographic/background information as opposed classifying them as a specific factor such as

social systems.

Items P7_Q1_02 and P7_Q1_03 parallel conflicts present in the clinical setting. Health

care institutions can provide social system support, such as continuing education for RNs.

Institutions have educational departments, direct access to nursing staff, and conduct ongoing

evaluation of patient care needs. However, health care institutions are also often the source of

barriers to gaining knowledge about genetics/genomics, negative attitudes, and decreased

adoption. These barriers are frequently related to time pressures such as poor staffing ratios and

work-related stressors. The GGNPS aims to comprehensively evaluate how genetics/genomics

are integrated into the clinical setting, including exploring knowledge/competency, confidence,

attitudes/receptivity, decision/adoption, as well as the role of social systems. However, each of

these components are individually complex. Particularly complex components, especially ones

where the relationship to other variables is not clear, such as social systems, should be further

evaluated in more focused analyses. Analyses may include focus groups to gain specific insight

into the relationship between health care institutions, nursing staff, facilitators/barriers to

continuing education, and which components encompass the variable of social systems. Such

focused analyses can further inform next steps and further analyses of the GGNPS.

126

Future analyses of construct validity. The EFA results shed light on why the CFA

analysis was unable to produce loadings. Some items designated as one variable in the GGNPS,

actually loaded on a different variable; as a result, the CFA was unable to identify factor loadings

based on the current GGNPS item designations.

Furthermore, some items in each variable (with the exception of confidence) actually

loaded on another variable, potentially resulting in measurement error. This measurement error is

compounded when evaluating direct and indirect relations among five variables. This provides

insight into many of the weak relations in the SEM; if the variables were measured incorrectly, as

indicated by the EFA, relations between the variables cannot be accurately evaluated. The SEM

should be reevaluated following revision of the variables.

In the SEM analysis of the original instrument, used by family physicians, Jenkins and

colleagues (2010) also noted that adoption of genetics/genomics did not follow the path predicted

by Rogers. Instead, there were two paths leading to adoption of genomic-related innovations:

one path based on ‘comfort’ with using genetics/genomics in clinical practice and a second path,

based on ‘relevance’ to clinical practice. The researchers hypothesized that the multiple paths

may be related to the complexity of the innovation, genetics/genomics, and complexity of clinical

practice itself. They contrasted this to the linear nature of Rogers’ DOI model. In clinical

practice, providers are influenced by a variety of factors, including the condition affecting the

patient and the provider’s approach to clinical practice. In the current study, the SEM results

demonstrated a lack of fit between the hypothesized model, based on Rogers’ DOI model, and

the data. Alternative paths will be explored in future analyses. However, prior to this, the EFA

results will be used to reclassify the variables, because many items did not load as predicted.

After reclassifying the variables, the SEM will be reevaluated. Alternative paths will be

127

considered by exploring both additional components of Rogers’ DOI model and paths related to

clinical practice, as reported by Jenkins and colleagues.

Future studies can also explore the effect of additional components of Rogers’ (2003)

DOI theory on adoption of genetics/genomics into nursing practice. For example, characteristics

of decision-making units and prior conditions. Although these components of Rogers’ theory are

not evaluated in the GGNPS, they can be studied using qualitative analyses to gain a focused

understanding of this portion of the theory. Specifically, innovativeness may be reflected by

adoption of other innovations, for example technologies such as electronic medicals records

(EMRs) and the adoption of innovations in one’s personal life. This would provide insight into

adoption of an innovation that is mandated by a nurse’s employer (e.g. EMR), as well as how

participants explore innovations on their own time. Felt needs/problems can be explored based

on nurses’ reports of what they need to effectively provide patient care and potential barriers in

the workplace. The researcher would attempt to identify themes related to how these

components of Rogers’ DOI theory affect adoption of genetics/genomics into nursing practice.

Evaluation of perceived characteristics of an innovation can provide additional insight

into barriers and facilitators of adopting genetics/genomics. These characteristics include relative

advantage, compatibility, complexity, trialability, and observability. The relative advantages of

using genetics/genomics are significant, for example allowing nurses to use the most up-to-date

clinical resources to guide screening, diagnosis, treatment selection, and health promotion. In

addition, because essentially all diseases have a genetic/genomic component, nurses are strongly

advantaged by having an understanding of this content and being able to use it in their clinical

practice (“FAQ About Genetic Disorders,” 2012). However, poor knowledge of

genetics/genomics may result in nurses not being aware of these advantages.

128

In addition, genetics/genomics may be incompatible with clinical institutions that do not

champion continued education about the topic, those that do not value the role of family history

evaluation, patient education about the genetic basis of disease, and/or identifying patients who

may benefit from referrals to genetic specialists. If the social system of the institution does not

emphasize and champion the role of genetics/genomics, it makes it increasingly difficult for

nurses to adopt it into their clinical practice. As a result, nurses will receive mixed messages

about patient care priorities. On one hand, nurses must administer medications and respond to

call bells in a timely manner, document carefully, and strive for high patient satisfaction, despite

poor nurse-to-patient ratios. On the other hand, institutions maintain that nurses must be current

in their clinical practice, although they may not provide the resources to help nurses learn about

and adopt clinical innovations. The resulting incompatibility may affect not only the feasibility

of adopting innovations, but also nurses’ attitudes/receptivity towards those innovations, as well

as their ability to increase their knowledge/competency and confidence toward the innovation.

Similarly, many aspects of genetics/genomics are complex; this complexity is a barrier to

the adoption of this content into nursing practice. The body of knowledge of genetics/genomics

is vast, information is frequently updated, and some concepts are intimidating for individuals

with no previous genetic/genomic experience. Also, adoption of genetics/genomics requires

clinical knowledge, associated skills (such as family history collection), and actions (such as

patient advocacy, patient education, and facilitating referrals).

Unlike some innovations that users can trial and evaluate prior to full adoption,

genetics/genomics have low trialability. Potential adopters cannot ‘try’ genetics/genomics like a

new computer and decide to adopt or reject it. This complexity, coupled with possible

incompatibility at some institutions, and the lack of ease of ‘trying’ the innovation before

adoption, makes genetics/genomics more difficult to adopt.

129

Not only are genetics/genomics more difficult to adopt, because it is an innovation that

affects so many aspects of patient care and has so many components, it is also difficult to observe

the adoption of this innovation. For example, nurses on a unit may see their colleagues using a

new piece of technology; they can easily observe this new technology/equipment in use and it

may promote the spread of an innovation in that social system. However, adoption of

genetics/genomics is not as easily observable. That affects the spread of the innovation among

their peers, other nurses may not recognize that their colleagues are using the innovation, and are

less likely to consider it for adoption to their practice. Decreased observability also makes

measuring adoption of genetics/genomics more difficult.

To overcome the difficulty of measuring adoption of genetics/genomics, it is important to

define how adoption is measured. Essential Competencies and Curricula Guidelines for Nurses

in Genetics and Genomics (2009) identify nursing roles and responsibilities required to adopt

genetics/genomics. These include collection and evaluation of family history information,

patient education, facilitating genetic referrals if warranted, and using genetic/genomic

information in clinical decision-making. All of these roles and responsibilities are reflected in

the GGNPS. Accurately evaluating adoption is essential because genetics/genomics affect how

we screen for, diagnose, and manage disease. In addition, genetics/genomics play a role in health

promotion and maintenance. Therefore, establishing validity of the GGNPS is essential to

measure adoption of genetics/genomics, assess a potential lack of adoption, and develop

strategies to promote the use of genetics and genomics in nursing practice to provide current,

safe, and effective patient care.

Implications for Nursing Research and Education

The GGNPS has a variety of applications for nursing education, practice, and research.

In nursing education, the GGNPS can be used to identify gaps in knowledge and create targeted

130

educational interventions. For example, it can be used to create continuing education initiatives

among practicing nurses and evaluate how those initiatives affect adoption of genetics/genomics

into nursing practice.

As the body of genetic/genomic knowledge rapidly expands, nursing curricula must also

incorporate this new information. Nursing programs increasingly focus on knowledge synthesis

and application to graduate nurses who can think critically and provide safe, competent patient

care. The GGNPS reflects needs of academic programs by evaluating both genetic/genomic

knowledge and applications of that content into nursing practice.

Genetics/genomics are pervasive in nursing practice, regardless of role or specialty.

Therefore, this content should be evaluated on licensing exams; however, integration of

genetics/genomics into licensing exams is a complex and ongoing process. New content is added

to licensing exams, such as the NCLEX, via multiple routes. Publications from the National

Council of State Boards of Nursing (NCSBN) (2016) state that the exam is updated with

evidence derived from clinical outcomes and by surveying newly licensed RNs in the first six

months of practice. Newly licensed RNs describe a list of activities based on their clinical

practice and this information is referenced for updating the NCLEX exam. However, because

integration of genetic/genomic content into nursing curricula is lagging, new graduates may not

be aware of their professional roles and responsibilities in relation to genetics/genomics or how

to use genetics/genomics in nursing practice. Similarly, new RNs are often task-based and

prioritize responsibilities such as medication administration and charting (Feng & Tsai, 2012).

As a result, because newly licensed RNs are less likely to consider the comprehensive roles and

responsibilities of a nurse, including genetic/genomic competencies, this content is less likely to

be integrated into NCLEX. Furthermore, a lack of questions on the licensing exam further

undermines the need to integrate genetics/genomics into nursing curricula. In addition, the lag in

131

integrating genetics/genomics into nursing curricula decreases its use in clinical practice; as a

result, less is known about how clinical outcomes are affected by genetics/genomics, which also

decreases its integration into content on licensing exams.

Evaluation of the NCLEX Detailed Test Plan (2016) references family history of

testicular cancer as a risk factor for increased risk of developing testicular cancer. Although that

information is correct, it is a very low level application of genetic/genomic knowledge. The

second reference to genetics/genomics is that performing a health history should include a

“genetic history” (National Council of State Boards of Nursing, 2016, p. 20). This is mentioned

in relation to health screening with no discussion of how to collect family history information for

multiple generations or use is it clinical decision-making. The current genetics/genomics content

in the NCLEX Test Plan is sparse and does not reflect the myriad of ways genetics/genomics

affects nursing practice and patient outcomes. The GGNPS can be incorporated into licensing

exam content to evaluate how genetic variation can affect disease processes and nurses' roles in

providing referrals, patient education, and advocacy.

Adoption of genetics/genomics into nursing practice is a complex process, affected by

many variables. In clinical practice, the GGNPS can be used to evaluate variables that are

facilitators and barriers to adoption of genetics/genomics. Evaluation of not only adoption, but

associated variables, is important because evidence indicates not only are genetics/genomics not

being fully adopted into clinical nursing practice, but inadequate knowledge, poor understanding

about relevance, inadequate confidence, and inconsistent support from social systems all

contribute to a lack of adoption (Calzone, Jenkins, Culp, Caskey, & Badzek, 2014; Calzone,

Jenkins, Culp, Bonham, & Badzek, 2013; Kirk, Lea, & Skirton, 2008).

A number of structural factors contribute to these barriers. For example, Calzone and

colleagues (2014) reported that following licensure, 86.8% (n = 4,594/5,294) of RNs

132

participating in a study did not complete any courses focusing on genetics/genomics. A lack of

learning opportunities directly contributes to inadequate knowledge about the topic. Social

system pressures are demonstrated by Sussner and colleagues (2011), inadequate time and a high

workload were reported as major barriers to collection of family history information. This can

also adversely affect nurses’ attitudes/receptivity; inadequate time and many additional

responsibilities may make nurses less likely to adopt genetics/genomics.

Additional structural barriers include a lack of a designated location to document family

history information and the inability to reference family history information that may have been

documented in the electronic health record (EHR) (Kho et al., 2011). As a result, nurses are

unsure of where to initially document the information or how to find family history information

that is located in an EHR. Barriers to adoption of genetics/genomics in clinical practice must be

addressed because genetics/genomics play an important role in both disease prevention and

health promotion. In managing clinical conditions, genetics/genomics affect screening,

diagnosis, treatment selection, and patient-education. Therefore, genetics/genomics are within the

scope of practice of all nurses, regardless of academic preparation, specialty, or clinical roles

(Jenkins & Calzone, 2007)

A lack of integration of genetics and genomics into nursing education and practice can be

addressed by championing nursing research. For example, future studies will use findings from

the EFA and feedback from content experts to restructure variables in the GGNPS and reevaluate

construct validity using both SEM and CFA analyses. Upon establishing face, content, and

construct validity, the GGNPS can be used to identify barriers to adoption of genetics/genomics,

develop initiatives to promote adoption, and use data from these initiatives to increases nurses’

knowledge, improve confidence, promote positive attitudes/receptivity, and increase social

system support. Additional focused, in-depth analyses can concentrate on complex factors such

133

as social systems. Analyses can focus on how health care institutions support the use of

genetics/genomics in nursing practice with evaluation of clinical sites who are early and effective

adopters. Additional focused studies can consider how decision-making units (e.g. peernetworks

in employment settings) affect nurses’ attitudes. Such focused initiatives can further inform

evaluation of the GGNPS. With continued evaluation of validity, the GGNPS can be used for a

variety of applications in nursing research, education, and clinical practice.

Appendix A

134

Diffusion of Innovations (DOI) Model

Appendix B

Knowledge Score Components

Item

Content

P2_Q3_01

“A family history that includes only 1st degree relatives such as

parents, siblings, and children should be taken for every new

patient.”

P2_Q3_02

“A family history that includes 2nd and 3rd degree relatives

such as grandparents, aunts, uncles, and cousins should be

taken for every new patient.”

P2_Q3_03

“Family history taking should be a key component of nursing

care.”

135

P2_Q3_04

“There is a role for nurses in counseling patients about

genetic risks.”

P4_Q1_01

“Do you think genetic risk has clinical relevance for: Breast

cancer.”

P4_Q1_02

“Do you think genetic risk has clinical relevance for: Colon

cancer.”

P4_Q1_03

“Do you think genetic risk has clinical relevance for:

Coronary heart disease.”

P4_Q1_04

“Do you think genetic risk has clinical relevance for:

Diabetes.”

Added after

revision

Extent to which family history supports clinical decisions

(such as administering drugs prescribed).

P5_Q1

“The DNA sequences of two randomly selected healthy

individuals of the same sex are 90-95% identical.”

P5_Q2

“Most common diseases such as diabetes and heart disease are

caused by a single gene variant.”

136

Appendix C.1

Q-Q Plot of Items Comprising the “Knowledge” Variable

Appendix C.2

Q-Q Plot of Items Comprising the “Attitude” Variable

Appendix C.3.

Q-Q Plot of Items Comprising the “Confidence” Variable

137

Appendix C.4.

Q-Q Plot of Items Comprising the “Social Systems” Variable

Appendix C.5.

Q-Q Plot of Items Comprising the “Adoption” Variable

138

139

Appendix D.

Welch-Two Sample T-Tests

Part 3, Question 4, Section 1 p

= 0.891

Confidence → Attitudes/Receptivity

.05

Part 1, Question 1 p

= 0.9769

Social Systems → Competency/knowledge .00205 Part 2, Question

3, Section 1 p = 0.7388

Part 2, Question 3, Section 2

p = 0.809

Part 2, Question 3, Section 3

p = 0.6151

Part 2, Question 3, Section 4

p = 0.8481

Part 4, Question 1, Section 1

p = 0.6694

Part 4, Question 1, Section 2

p = 0.7097

Part 4, Question 1, Section 3

p = 0.7077

Part 4, Question 1, Section 4

p = 0.7691

Part 4, Question 1, Section 5

p = 0.5768

Part 4, Question 1, Section 6

p = 0.9776

Variables

Family

wise

Alpha value

value

Confidence

→

Decision/Adoption

.0127

Part 3, Question 2

0.268

Part 3, Question 3

0.338

Part 3, Question 4, Section 1

0.593

140

Part 4, Question 2, Section 1

p = 0.3955

Part 4, Question 2, Section 2

p = 0.5243

Part 4, Question 2, Section 3

p = 0.321

Part 4, Question 2, Section 4

p = 0.529

Part 4, Question 2, Section 5

141

p = 0.6829

Part 4, Question 4, Section 1

p = 0.4489

Part 4, Question 4, Section 2

p = 0.4771

Part 4, Question 4, Section 3

p = 0.07002

Part 4, Question 4, Section 4

p = 0.1501

Part 4, Question 4, Section 5

p = 0.449

Part 4, Question 4, Section 6

p = 0.5425

Part 5, Question 1

p = 0.9821

Part 5, Question 2

p = 0.9327

Part 6, Question 1

p = 0.7105

142

Part 3, Question 3 p

143

Part 6, Question 2, Section 1

p = 0.8404

p = 0.02784 Part 3, Question 3

p = 0.02431

Part 3, Question 4, Section 1

p = 0.02998

Part 3, Question 4, Section 2 p

= 0.008711

Social Systems → Attitudes/Receptivity

.05

Part 1, Question 1 p

= 0.7577

Attitudes/Receptivity → .127 Part 3, Question 2

Decision/Adoption p < 0.001

Part 3, Question 3

p < 0.001

Part 3, Question 4, Section 1

p < 0.001

Competency/knowledge → .127 Part 3, Question 2

Decision/Adoption p < 0.001

Part 3, Question 4, Part 1 p

< 0.001

Part 2 Question 1, Section 2

p = 0.9102

Part 2, Question 1, Section 3

p = 0.9806

< 0.001

Part 6, Question 2, Section 2

=0.7472

Social Systems

→

Decision/Adoption

.127

Part 3, Question 2

Part 3, Question 4, Section 2

Part 3, Question 4, Part 2

0.001

ompetency/knowledge

→

Confidence

.0366

Part 2 Question 1, Section 1

0.999

144

Part 2, Question 1, Section 4

p = 0.6327

Part 2, Question 1, Section 5

p = 0.5057

Part 2, Question 1, Section 6

p = 0.4773

Part 2, Question 1, Section 7

p = 0.8218

Part 2, Question 1, Section 8

p = 0.5948

Part 2, Question 1, Section 9

p = 0.7042

Part 2, Question 1, Section

10 p = 0.8568

Part 2, Question 1, Section

11 p = 0.8496

Part 2, Question 1, Section

12 p = 0.7663

Part 2, Question 1, Section

13 p = 0.6453

Part 7, Question 1, Section 2

p < 0.001

Part 7, Question 1, Section 3

p < 0.001

Part 7, Question 1, Section 4

p < 0.001

Part 2, Question 1, Section 14

0.753

Decision/Adoption

→

Social Systems

.102

Part 7, Question 1, Section 1

0.001

145

Part 7, Question 1, Section 5 p

< 0.001

Decision/Adoption → Attitudes/Receptivity

.0032

Part 1, Question 1 p

< 0.001

Decision/Adoption →

.00366

Part 2, Question 1, Section 1

Competency/knowledge p < 0.001

Part 2, Question 1, Section 2

p < 0.001

Part 2, Question 1, Section 3

p < 0.001

Part 2, Question 1, Section 4

p < 0.001

Part 2, Question 1, Section 5

p < 0.001

Part 2, Question 1, Section 6

p < 0.001

Part 2, Question 1, Section 7

p < 0.001

Part 2, Question 1, Section 8

p < 0.001

Part 2, Question 1, Section 9

p < 0.001

Part 2, Question 1, Section

10 p < 0.001

Part 2, Question 1, Section

11 p < 0.001

Part 2, Question 1, Section

12 p < 0.001

Part 2, Question 1, Section

13 p < 0.001

146

Part 2, Question 1, Section 14

0.001

147

Appendix E

Content Validity Index and Ratio

Domain: Attitudes/Receptivity

Conceptual definition: “The relative advantage offered by the innovation. Perceived attributes of the

innovation including the recognized need for the innovation, as well as contextual factors such as

identification of communication channels that facilitate or hamper adoption and sustained

implementation” (Calzone et al., 2012, p. 12).

Contextual definition: “Perceived importance, advantages, and disadvantages of integrating genomics

into practice. The complexity of integrating family history into practice. Perception of the value of

family history in patient care and their personal practice” (Calzone et al., 2012, p. 12).

1. Please indicate whether you think each of the following

would be a potential disadvantage of integrating genetics of

common diseases into your practice:

Relevancy (1-4)

1- Not

relevant

2- Somewhat

relevant

3- Relevant

4- Very

relevant

CVR

0.714

148

* Frequency of responses

Domain: Attitudes/Receptivity cont.

2. Please indicate whether you think each of the following

would be a potential advantage of integrating genetics of

common diseases into your practice:

3. How important do you think it is for the nurse to become

more educated about the genetics of common diseases?

 Very important  Somewhat important  Not very

important  Not at all important  Neutral

 Don’t know

Relevancy (1-4)

1- Not

relevant

2- Somewhat

relevant

3- Relevant

4- Very

relevant

CVR

0.714

Relevancy (1-4)

1- Not

relevant

2- Somewhat

relevant

3- Relevant

4- Very

relevant

CVR

No Advantage

Advantage

Better treatment

decisions (e.g., which

drugs to prescribe)

Improved services to

the patients

Better adherence to

clinical

recommendations

among patients

149

4. Please indicate whether you agree or disagree with the

following statements.

Domain: Confidence

Conceptual definition: “Level of certainty that knowledge about the innovation is accurate” (Calzone et

al., 2012, p. 12).

Contextual definition: “Confidence in: discussing genetics with patients; deciding what family history

information is relevant to assessing genetic susceptibility; availability, risks, benefits and limitations of

genetic testing; facilitating referral for genetic services” (Calzone et al., 2012, p. 12).

Relevancy (1-4)

1- Not

relevant

2- Somewhat

relevant

3- Relevant

4- Very

relevant

CVR

Relevancy (1-4)

1- Not

relevant

2- Somewhat

relevant

3- Relevant

4- Very

relevant

CVR

Agree

Disagree

Don't Know

Family

history taking

should be a

key

component of

nursing care.

150

5. Each of the following statements relates to the genetics of

common diseases and family history taking. By common

diseases, we are referring to disorders that arise as a result

of interactions between an individual’s environment and

his or her unique genetic makeup. Common diseases include

diseases such as cancer, heart disease, and diabetes. Please

indicate how confident you are that you can do each of

the following:

Decide what family history information is needed to tell

something about a patient’s genetic susceptibility to

common diseases.

Discuss how family history affects recommended screening

intervals.

Decide which patients would benefit from a referral for

genetic counseling and possible testing for susceptibility to

common diseases

Access reliable and current information about genetics and

common diseases.

Give patients information about the risks of genetic testing

for common diseases

Give patients information about the benefits of genetic

testing for common diseases.

Give patients information about the limitations of genetic

testing for common diseases.

Facilitate referrals for genetic services for common

diseases.

Not at all Confident confident

Domain: Competency/ Knowledge

Conceptual definition: “Recognition of the innovation and evidence of understanding the innovation and

its function” (Calzone et al., 2012, p. 12).

Contextual definition: “Knowledge of the genomics of common disease, and the family history

information needed to evaluate patients’ genetic susceptibility” (Calzone et al., 2012, p. 12).

0.714

0.429

0.714

151

6. Please indicate whether you agree or disagree with the

following statements.

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

0.714

152

7. Do you think that genetic risk (e.g., as indicated by family

history) has clinical relevance for the following:

Domain: Competency/ Knowledge cont.

8. When patients indicate a disorder in the family, which of

the following pieces of information do you collect in your

standard family history assessment? Each family member’s:

** One reviewer left answer blank

9. Thinking about how you support clinical decisions (such as

administering drugs prescribed), how important do you think

each of the following is to consider?

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

0.714

6**

0.714

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

5**

0.429

0.714

Not at

all

Essential

Don't know

Genetic Test Result

Family history

153

10. The DNA sequences of two randomly selected healthy

individuals of the same sex are 90-95% identical.

 true  false  don't know

11. Most common diseases such as diabetes and heart disease

are caused by a single gene variant.

 true  false  don't know Domain:

Competency/ Knowledge cont.

12. The Essential Competencies and Curricula Guidelines for

Nurses in Genetics and Genomics are endorsed as being a

standard part of nursing practice. Have you heard or read

about these Competencies?

 Yes

 No

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

3**

0.428

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

1**

-0.714

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

4**

0.714

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

0.714

13.

Excellent

Please rate your

understanding of the

genetics of common

diseases

In describing your

genetic/genomic

knowledge, would you

consider it to be

154

Domain: Social System

Conceptual definition: “The environment in which the innovation is being introduced, the

interconnected entities involved in shared problem solving” (Calzone et al., 2012, p. 12).

Contextual definition: Supervisory support for nurses using genomics, and institutional financial support

for genomic continuing education ” (Calzone et al., 2012, p. 12).

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

0.142

0.714

Domain: Decision/Adoption

Conceptual definition: “Observation of use of the innovation” (Calzone et al., 2012, p. 12).

155

Contextual definition: “Utilization of family history within the past three months” (Calzone et al., 2012,

p.12). Use of family history was selected as a measure of adoption because it is independent of

availability of technology and financial resources.

15. In the past three months, how often have you collected a

complete family history from a patient that includes the

following components: information on disorders from three

generations, and age at diagnosis and death for each affected

family member?

 Always  Often  Occasionally  Rarely or Never

16. In the past three months, has any patient initiated a

discussion with you about genetics?

 Yes

 No

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

156

17. Thinking specifically about patients that you have seen in

the past three months, please answer the following

questions.

Relevancy (1-4)

1- Not relevant

2- Somewhat relevant

3- Relevant

4- Very relevant

CVR

6**

0.714

157

Appendix

158

F.1

Path Model Using Imputation 1

This path model depicts the relations among variables in the analysis. p* < 0.001; p** =

0.077

F.2

Path Model Using Imputation 2

Social

System

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

355

0.260

.622

0.026

.062

0.323

0.344

Appendix

159

This path model depicts the relations among variables in the analysis. p*< 0.001; p** =

0.017

F.3

Path Model Using Imputation 3

Social

System

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

350

0.280

.659

.343

0.035

.093

0.325

Appendix

160

This path model depicts the relations among variables in the analysis. p*< 0.001; p** =

0.002

F.4

Path Model Using Imputation 4

Social

System

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

337

0.255

.637

.337

0.046

.072

0.325

Appendix

161

This path model depicts the relations among variables in the analysis. p*< 0.001; p** =

0.010

β weight Standardization

Figure 10

SEM β weight Standardization Original Data Set.

Social

System

Competency

/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

372

0.252

.628

.336

0.037

.079

0.325

Appendix

162

p* < .001; p** = .013

Although we were unable to arrive at a stable set of model parameters for the CFA, the

final model parameters achieved through the CFA do provide some insight into the relation

between the theoretical latent variables. The standardized root mean square residual (SRMR),

an indication of how well the model fit the data was .062; Hu & Bentler (1998) suggest that

SRMRs of less than .08 indicate an adequate fit the data to the model. Therefore, although we

cannot

Social Systems

Competency/Knowledge

Attitudes/Receptivity

Decision/Adoption

Confidence

.300*

.101*

.407*

.449*

.041**

.123*

.258*

163

regard the last model computed through the iterative CFA to represent a reliable or definitive set

of model parameters, by conventional standards of structural equation modeling, the model

presents sufficient fit that we may tentatively explore its parameters for insights into the

relations between the proposed latent variables.

Figure 11 presents the β weights between the standardized latent variable scores. We can

review the weights in the model presented in Figure 11 by starting with the strongest relations.

Beta weights indicate the amount of change in a given standardized ‘outcome’ variable for every

standardized unit in the predictor variable(Cribari-Neto & Zeileis, 2009). The effect of

competency/knowledge on confidence had the largest β weight (.449, p <.001). In other words,

for every one-point change in the standardized competency/knowledge score, which is one

standard deviation of change, there is a .449 point change in confidence. Indicating that an

increase in ones’ competency/knowledge accounted for a substantial portion in the increase in

confidence.

The second largest effect was that of social systems on competency/knowledge (β =

.407, p < .001). Social system support significantly affected nurses’ competency/knowledge in

the hypothesized model. In addition, there is a strong effect of social systems on

decision/adoption (β = .300, p <. 001). However, effects of the social system variable must be

cautiously interpreted, because it consists of only five items. Furthermore, the EFA

demonstrated that most of the social system items may be related to other variables and the

method of interpreting social systems effects should be reevaluated in future studies.

Surprisingly, the hypothesized model shows a negative effect of attitude/receptivity on

decision/adoption (β = -.258, p <. 001). By contrast, the literature review describes positive

Students also viewed