Attitudes towards genetics and genetic testing among participants in the Jackson and Framingham Heart Studies

Abstract

Genetic analysis has become integral to many large cohort studies. However, little is known about longitudinal cohort study participants’ attitudes towards genetics and genetic testing. We analyzed data from a survey of participants in the Jackson Heart Study (n=960), Framingham Heart Study (n=955), and Framingham Heart Study-Omni Cohort (n=160). Based on a three-question attitude scale, most participants had positive attitudes towards genetic testing (median score 4.3–5/5). Participants were also asked to select words to describe their attitudes towards genetics. More respondents endorsed the positive words “hopeful” (60–70%), “optimistic” (44–64%), “enthusiastic” (35–43%), or “excited” (28–30%) than the negative words “cautious” (35–38%), “concerned” (25–55%), “worried” (6–13%), “pessimistic” (2–5%) or “horrified” (1–5%). Characteristics associated with favorable attitudes were greater genetics knowledge, higher subjective numeracy, experience with genetic testing, less frequent religious attendance, and not being employed. These findings demonstrate variation in attitudes even among participants in long-standing cohort studies, indicating a need for ongoing participant engagement and education.

INTRODUCTION

Pairing genetic and genomic analyses with clinical and environmental data in large, long-term prospective research projects promises to expand our knowledge of human health and disease. This integration of multiple types of health-related data is the focus of high-priority national initiatives such as the Precision Medicine Initiative’s All of Us program. The success of these initiatives depends on participants’ long-term engagement in and support for research activities. However, little is known about what influences variation in attitudes towards genetics and genetic testing among participants in long-standing cohort studies.

Public attitudes towards genetic testing are generally positive compared to other uses of biotechnology and have become more positive over time (Condit, 2010; Haga et al., 2013; Henneman et al., 2013). Most people are enthusiastic about genetic testing that would lead to actionable results (including reproductive decision-making), though only about a quarter support testing newborns or children for adult-onset conditions (Chokoshvili et al., 2017; Vermeulen, Henneman, van El, & Cornel, 2014).

A number of studies have shown that respondent characteristics are associated with variation in attitudes and that these associations are conditional on the purpose of testing. In general, younger people, people with higher educational attainment, and people with personal experience of genetic testing have more positive attitudes towards genetic testing (Etchegary et al., 2010). Higher educational attainment and higher socioeconomic status are associated with more positive parental attitudes towards pediatric diagnostic testing (Lim et al., 2017); whereas lower educational attainment is associated with greater interest in genetic testing of newborns for adult onset conditions and greater interest in predictive genetic testing in adults (Chokoshvili et al., 2017; Vermeulen et al., 2014).

The relationship between attitudes and respondent genetic literacy is similarly complex. Many studies report that greater genetic literacy among the general public is associated with more positive attitudes towards predictive genetic testing and genetics generally (Rose, Peters, Shea, & Armstrong, 2005; Sanderson, Wardle, & Michie, 2005). Other studies, however, have found that greater genetic literacy is associated with greater reported fear and uncertainty about the implications of genetic testing and less support for morally contentious uses of genetic testing (Etchegary et al., 2010; Haga et al., 2013). This suggests that more knowledgeable participants are attentive to the nuanced benefits, limitations, and risks of genetic testing.

Attitudes also vary by respondent race or ethnicity, and religious involvement. Members of racial and ethnic minority groups report generally positive attitudes towards clinical testing and are interested in contributing to knowledge relevant to their communities, but they also have greater concerns about stigma, discrimination, misuse of genetic information and practical barriers to testing than members of other groups (Dye et al., 2016; Hamilton et al., 2016; Hann et al., 2017; Lewis et al., 2018; Peters, Rose, & Armstrong, 2004). Additionally, greater religious involvement has been found to be associated with more negative attitudes towards genetic testing (Botoseneanu, Alexander, & Banaszak-Holl, 2011; Chokoshvili et al., 2017).

General attitudes towards genetics and genetic testing are closely tied to attitudes about joining genetics research studies, including contributing biospecimens or DNA to research biobanks. Surveys of the public on hypothetical precision medicine or biobank studies have found that 54%−66% of respondents definitely or probably would participate in a study if asked. Greater familiarity with genetics and genomics, personal experience with genetics, younger age, higher educational attainment, lower religiosity, and identifying as white are all associated with willingness to contribute DNA and medical data to hypothetical research studies (Dye et al., 2016; Kaufman, Baker, Milner, Devaney, & Hudson, 2016; Middleton et al., 2018; Sanderson et al., 2017). Excessive skepticism and concerns about the potential harms of genetic analysis can hinder recruitment and slow research progress. Limited evidence suggests that people with more negative attitudes towards genetics may be less likely to participate in research (Halbert, Gandy, Collier, & Shaker, 2006).

Both positive and negative attitudes towards genetics and genetic testing could have implications for research participants’ satisfaction with and commitment to participation once they have enrolled. Excessive enthusiasm among enrolled participants, potentially fueled by direct-to-consumer marketing and headlines about scientific breakthroughs, could foster unreasonable expectations about the promise of genetics research and disappointment if expectations are not met. Among participants of African descent in a sequencing study, nearly a third held unrealistic expectations that genomic sequencing could provide them with a clean bill of health (Lewis et al., 2018). In a sequencing study of healthy adults and cardiomyopathy patients, research participants experienced less clinical utility from exome sequencing than they expected prior to getting their results (Roberts et al., 2018). There is little direct data on how unrealistic expectations might play into attitudes towards genetics or satisfaction with ongoing participation.

This body of research provides information on the general public’s attitudes towards genetics and the factors associated with willingness to join genetics research studies. Knowing what factors are associated with positive or negative attitudes towards genetics and genetic testing among current research participants is essential to developing successful research programs that can recruit and retain participants who “reflect the rich diversity of the U.S.” (“PMI Cohort Program announces new name,” 2016). Additional insight can be gained from understanding variation in attitudes towards genetics and genetic among participants in long-term natural history studies, such as the Jackson Heart Study (JHS, established in 1998) and the Framingham Heart Study (FHS, established in 1948, and augmented with the more diverse FHS-Omni cohort in 1994).

JHS and FHS are community-based, longitudinal cohort studies funded by the National Heart, Lung, and Blood Institute (NHLBI) to study cardiovascular disease incidence, natural history, and risk factors in defined populations. These studies have contributed to many arenas of biomedical and social science, including but not limited to cardiovascular disease. JHS and FHS cohort participants have demonstrated exceptional long-term commitment to research and therefore represent a critical scientific resource. Although the primary focus of JHS and FHS is cardiovascular health, pre-existing cardiovascular disease is not an eligibility requirement for enrollment, and these studies conduct secondary research related to other areas of health and illness. They are, therefore, more similar to general cohorts than to disease-specific cohorts. Though JHS and FHS were not primarily designed as genetics research studies, familial inheritance and genetic analysis are integral study components. JHS and FHS began collecting and analyzing DNA from some participants in 1998. In 2005, NHLBI launched an effort to conduct genome-wide association studies (GWAS), and later whole-genome sequencing, in existing large cohorts including JHS and FHS. Eighty-seven percent of JHS participants and over 99% of FHS participants have consented to genetic analysis within their parent studies (Wilson et al., 2005; Levy et al., 2010).

JHS and FHS cohorts serve as a model for participants in other similar research programs, so understanding their attitudes can inform future efforts for participant recruitment and engagement. To augment the limited data addressing attitudes among current research participants, we analyzed experiential, knowledge and demographic factors associated with having more positive or more negative attitudes towards genetics and genetic testing among participants in JHS and FHS as well as African-American members of the FHS-Omni cohort.

MATERIALS AND METHODS

Participants

The study reported here is a secondary analysis of data from a survey designed to assess preferences for learning individual genetic testing results (return of results) among members of the JHS, FHS, and FHS-Omni cohorts who had previously consented to participate in the genetic components of their respective studies.

The Jackson Heart Study (JHS)

An expansion of the Jackson Field Center of the Atherosclerosis Risk in Communities (ARIC) Study, JHS was launched in 1998 to study cardiovascular disease risk factors, long-term trends and prevention in African-American men and women (Fuqua et al., 2005). JHS enrolled over 5300 participants in the Jackson, MS metropolitan area, recruited randomly and enriched with families with at least three first-degree relatives willing to participate. In response to greater recruitment and retention challenges encountered in the all-African-American Jackson ARIC site compared to other ARIC sites, JHS prioritized participant education and implemented a community-based participatory research approach to study planning (Walker et al., 2014; Wyatt et al., 2003). Participants have completed three clinical exams to date; a fourth exam is planned for 2020–2022. Participants were given the option to consent to genetic analysis (described as candidate gene analysis, GWAS, and other genomic analyses) and to specify whether their data could be used by non-JHS researchers or for research questions beyond cardiovascular disease. To inform participants about the genetic analysis components of the study, JHS conducted public lectures and disseminated written educational materials (Wilson et al., 2005). Eighty-seven percent of the total cohort gave consent for genetic analysis and data sharing with JHS collaborators. For the present study, surveys were mailed to a randomly selected subset (n=1426) of the 4615 participants who had consented to genetics research conducted by JHS collaborating investigators.

Framingham Heart Study (FHS)

The FHS was initiated in 1948 to study the natural progression, risk factors, and prognosis of cardiovascular, respiratory and other diseases. It is among the most comprehensive and longest-running health studies in the world. The FHS prioritized enrollment of family members with large pedigrees, and has achieved 99% retention of participants over many years of regular exam visits (Tsao & Vasan, 2015). There have been three generations of FHS participants: the Original cohort starting in 1948 (n=5209), the Offspring cohort starting in 1971 (n=5124) (plus later addition of the Offspring Spouse cohort, n=103), and the Third Generation cohort starting in 2002 (n=4095). Participants were informed of planned genetics studies through newsletters and consent discussions at clinic visits and were given options to opt in or out of various types of genetic analysis and data use. Over 99% of Offspring and Third Generation cohort participants consented to DNA extraction, genetic data sharing with researchers, cell line creation, and research on potentially sensitive topics (Levy et al., 2010). As described in depth in Levy et al., Offspring Cohort participants provided this consent in their eighth clinic visits; Third Generation Cohort participants provided this consent in their first and second clinic visits, and consent preferences are updated with each clinic visit. For the present study, surveys were mailed to a randomly selected subset (n=1308) of the over 7000 FHS Offspring and Third Generation participants who had consented to participate in genetics research and data sharing with qualified researchers.

Framingham Heart Study – Omni Cohort (FHS-Omni)

FHS participants are predominantly white. In recognition of increasing diversity in the Framingham region, the FHS added the smaller Omni cohorts, enrolling African-American, Hispanic, Asian, Indian, Pacific Islander and Native American residents of Framingham and surrounding towns, in 1994 (n=506) and 2003 (n=410). Consent form options and exam protocols were harmonized with the main FHS cohorts in 1999. Surveys were mailed to all 226 self-identified African-American (including mixed-race) members of the Omni cohorts who had previously consented to participate in genetics research and data sharing with qualified researchers.

Survey instrument, main outcomes and covariates

Two measures of attitudes towards genetics and genetic testing were adapted from existing measures (Michie, di Lorenzo, Lane, Armstrong, & Sanderson, 2004; Sanderson et al., 2005). For the genetic testing attitude scale, participants were asked to rate three items on a scale from one to five. The survey prompt reads: “For each of the following items, please check the box that best describes your attitude about having a genetic test.” “A bad thing/A good thing,” “Beneficial/Harmful,” and “Important/Not important.” The measure is scored by averaging across the three items, after reverse-scoring the 2^nd and 3^rd items. Internal reliability (Cronbach’s alpha) for the three items was 0.80, 0.87 and 0.83 in the JHS, FHS and FHS-Omni samples, respectively. Because most participants had very positive attitudes (median scores of 5.0, 4.7 and 4.3 for JHS, FHS and FHS-Omni), we chose to report factors associated with scoring in the bottom (most negative) quartile of respondents rather than the inverse (i.e., reporting factors associated with more positive scores). Responses were dichotomized into the top 74.1% (scores of 4.00 and above) and the bottom 25.9% (scores of 3.67 and below) of the combined study sample.

The genetics attitude measures, positive word score and negative word score, are based on a 12-item attitude word list. The survey prompt reads: “Below is a list of words that describe how some people feel about genetics. Please read the list of words. Then check all the words that describe how you feel about genetics. You may check as many or as few words as you wish.” Positive words included enthusiastic, optimistic, hopeful, and excited; ambivalent words included indifferent, confused, and mixed feelings; negative words included horrified, concerned, pessimistic, worried, and cautious. In previous studies, the positive words loaded together, as did the negative words, and the two sets of words were independent of each other (Sanderson et al., 2005), so we examined them as distinct measures. In the current study, we dichotomized high scores on either measure as close to the top quartile as the data allowed. High positive word score is defined as 3 or more positive words (the most positive 33.2% of participants). High negative word score is defined as 2 or more negative words (the most negative 25.6% of participants).

Knowledge of genetics and genetic testing was measured by counting the number of correct responses to a measure combining four questions from Singer et al. and three questions from Furr et al. (Furr & Kelly, 1999; Singer, Antonucci, & van Hoewyk, 2004). The following true or false statements were included in the final measure (correct answers in brackets): “1. If a person has a genetic mutation for a disease, the person will always get the disease [F]. 2. Only mothers can pass on genetic diseases [F]. 3. People can be healthy even if they have a genetic mutation for a disease [T]. 4. Genetic testing can be used in adults to find out if they have a greater than average chance of developing certain kinds of cancer [T]. 5. Genetic testing can be used in adults to find out if they have a greater than average chance of developing depression [F]. 6. Genetic testing can be used in adults to predict whether a person will have a heart attack [F]. 7. Genetic testing can be used during pregnancy to find out whether the baby will develop sickle cell disease or cystic fibrosis [T].”

A single item was used to measure personal experience with genetic testing: “Have you ever had a genetic test to find out if you are at increased risk for a disease?”; “no” and “not sure” were combined in analysis. Respondents rated their subjective numeracy using four items from Fagerlin et al.’s Subjective Numeracy Scale (SNS) that asked participants to rate how comfortable they are using fractions and percentages, how helpful they find tables and graphs, and whether they prefer numerical information or narrative information (Fagerlin et al., 2007). In pilot testing, a scale based on four of the eight items was highly correlated with the full eight-item scale (r=0.95), so respondent burden was reduced by using only four items.

Demographic questions included educational attainment (categories collapsed to “college graduate and above” and “below college graduate” for analysis), work status (categories collapsed to “employed” and “not employed” for analysis), marital status, whether the respondent had any children, religious preference, and how frequently the respondent attended religious services. FHS and JHS provided the age and sex of each respondent.

Survey administration

Self-administered paper surveys were mailed by the JHS and FHS coordinating centers to potential participants in late 2011 and early 2012 and returned by mail. Potential participants received up to five mailings to maximize response rates: a pre-notification letter, the first survey mailing, a follow-up to the first survey mailing, and two additional duplicate survey mailings. All materials used standard JHS and FHS branding and letterhead.

Statistical analysis

Regression analyses were carried out to test specific hypotheses of the associations between participant characteristics and the three attitude measures: more negative attitude, high positive word score, and high negative word score.

Based on previous studies, we hypothesized that positive attitudes would be associated with greater knowledge of genetics and genetic testing, higher educational attainment, higher subjective numeracy, personal experience with genetic testing, being female, and having less religious involvement. Additional covariates for which we did not have hypotheses based on prior literature were work status, marital status, having children, and religious preference. As pre-specified in the study protocol, analyses were conducted separately within each cohort, not on the combined population. As a result, we could not test hypotheses regarding race or ethnicity because, by design, there was little within-cohort racial or ethnic diversity. No statistical comparisons were made across cohorts.

For each outcome measure, a multi-step process was used to build logistic regression models. First, bivariate associations between each outcome measure and each covariate were tested using chi-square, t- or Wilcoxon rank-sum tests to identify participant characteristics that may be associated with the three outcome measures. All tests were two-sided. The following covariates were tested: genetic knowledge, personal experience with genetic testing, subjective numeracy, educational attainment, work status, participant age, sex, marital status, having children, and frequency of religious attendance. Religious preference was not included because the group sizes were too small. Covariates that met the pre-defined threshold of p<0.2 were then included in model building. See Supplemental Table S1 for the results of bivariate analysis for variables that passed the p<0.2 threshold to be included in subsequent model building.

Next, we ran logistic regression models with all possible combinations of covariates (depending on which covariates met the criteria for inclusion for each cohort and each outcome measure). The models with the lowest Akaike information criterion (AIC) and the lowest Bayesian information criterion (BIC) were identified, and covariates retained in both of these models were included in final models (Kuha, 2004). Any variables in the final models with p<0.05 were considered to have significant associations with the attitude outcome variables.

Post-hoc power analysis

Sample sizes for JHS, FHS and FHS-Omni after removing individuals with missing data were 882, 911 and 148 for the negative attitude scale, 877, 929 and 143 for the positive word scale, and 913, 876 and 152 for negative word score. Post-hoc power calculations are based on n=900 for JHS and FHS, and n=150 for FHS-Omni, using two-means power analysis.

The JHS and FHS samples are well powered to detect relatively small effect sizes for most variables. For covariates with fairly even distribution of responses across categories (no greater than 2/1 difference between categories), a small effect size of 0.2 standard deviations could be detected with 81% power (α=0.05). For covariates where some responses were chosen by only 1/10 of participants, a medium effect size of 0.5 standard deviations could be detected with 99% power.

The smaller FHS-Omni sample is powered to detect medium to large effects. For covariates with a 2/1 response ratio, a medium effect size of 0.5 standard deviations could be detected with 82% power. For variables with a 10/1 response ratio, a large effect size of 0.8 standard deviations could be detected with 78% power.

RESULTS

Description of the study samples

Survey response rates were 67.4%, 73.0%, and 70.8% for JHS, FHS and FHS-Omni, respectively. There were no statistically significant differences in age or sex between JHS respondents and non-respondents (data not shown). FHS respondents were slightly older than non-respondents (mean age 58.3 vs. 55.9), but did not differ by sex. Omni respondents were slightly older than non-respondents (mean age 64.1 vs. 60.4) but did not differ by sex.

Table 1 reports the demographic and attitudinal characteristics of the respondents. Personal experience with genetic testing was uncommon; 13.4% (JHS), 8.2% (FHS) and 5.7% (FHS-Omni) reported having had a prior clinical genetic test. Knowledge of genetics and genetic testing was moderate, with mean scores of 3.5, 4.1 and 3.8 out of 7. Subjective numeracy was also moderate, with mean responses of 3.5, 4.3 and 4.1 out of 6 for JHS, FHS and FHS-Omni respondents respectively. There was substantial variation in educational attainment, genetic literacy and subjective numeracy within each cohort.

Table 1.

Respondent characteristics by cohort.

Characteristic	JHS (n = 960)	FHS (n = 955)	FHS Omni (n = 160)
Age, Valid n (% of total)	956 (99.6%)	955 (100%)	160 (100%)
Mean Age (standard deviation)	63.2 (12.1)	58.3 (13.9)	64.1 (13.0)
Median Age (min-max)	63 (29–95)	58 (27–93)	65 (27–89)
Sex, Valid n (% of total)	956 (99.6%)	955 (100%)	160 (100%)
% Female	66.2%	55.2%	61.9%
Race, Valid n (% of total)a	960 (100%)	931 (97.5%)	155 (96.9%)
African-American	100%	0%	88.4%
White	0%	98.7%	0.6%
Otherb	0%	1.3%	11.0%
Ethnicity, Valid n (% of total)	879 (91.6%)	928 (97.2%)	149 (93.1%)
Hispanic	2.5%	1.9%	0.7%
Education, Valid n (% of total)	921 (95.9%)	947 (99.2%)	154 (96.3%)
High school graduate or less	31.4%	18.2%	9.1%
Some college or technical school	22.7%	21.1%	15.6%
College graduate or higher	45.9%	60.7%	75.3%
Employment Status, Valid n (% of total)	927 (96.6%)	939 (98.3%)	156 (97.5%)
Employed	41.9%	59.0%	43.6%
Not employedc	58.1%	41.0%	46.4%
Marital Status, Valid n (% of total)	913 (95.1%)	940 (98.4%)	151 (94.4%)
Single, never married	11.5%	7.9%	10.6%
Married or living with partner	48.5%	75.0%	55.6%
Divorced or separated	20.4%	10.3%	19.9%
Widowed	19.6%	6.8%	13.9%
Has children, Valid n (% of total)	930 (96.9%)	943 (98.7%)	157 (98.1%)
Yes	89.6%	80.5%	86.0%
Religious attendance, Valid n (% of total)	901 (93.9%)	888 (93.0%)	151 (94.4%)
Nearly every day	19.6%	2.6%	9.3%
At least once a week	59.6%	20.4%	43.7%
Few times a month	14.1%	12.2%	13.9%
Few times a year	4.8%	27.1%	13.9%
Less than once a year or not at all	1.9%	37.7%	23.2%
Religious preference	-	-	-
Protestant, Baptist	59.4%	4.7%	29.4%
Protestant, Mainline	13.2%	10.0%	22.5%
Protestant, other	5.1%	6.2%	11.3%
Catholic	2.5%	51.4%	8.1%
Other	11.4%	8.1%	10.0%
None	1.4%	12.7%	12.5%
Prefer not to state	4.4%	5.8%	3.8%
No response	2.7%	1.1%	2.5%
Personal experience with genetic testing, Valid n (% of total)	941 (98.0%)	939 (98.3%)	157 (98.1%)
Yes	13.4%	8.2%	5.7%
Knowledge of genetics and genetic testing, Valid n (% of total)d	936 (97.5%)	937 (98.1%)	157 (98.1%)
Mean (standard deviation)	3.5 (1.7)	4.1 (1.5)	3.8 (1.7)
Median (25th, 75th percentile)	4 (2, 5)	4 (3, 5)	4 (3, 5)
Subjective numeracy, Valid n (% of total)e	944 (98.3%)	945 (99.0%)	159 (99.4%)
Mean (standard deviation)	3.5 (1.3)	4.3 (1.2)	4.1 (1.3)
Median (25th, 75th percentile)	3.7 (2.5, 4.5)	4.5 (3.5, 5.3)	4.3 (3, 5.3)
Attitude toward genetic testing, Valid n (% of total)f	935 (97.4%)	935 (97.9%)	157 (98.1%)
Mean (standard deviation)	4.4 (0.9)	4.4 (0.8)	4.1 (0.9)
Median (25th, 75th percentile)	5.0 (3.7, 5.0)	4.7 (4.0, 5.0)	4.3 (3.3, 5.0)
Number of positive words endorsed, Valid n (% of total)g	945 (98.4%)	947 (99.2%)	159 (99.4%)
Mean (standard deviation)	1.7 (1.3)	2.1 (1.3)	1.9 (1.3)
Median (25th, 75th percentile)	2 (1, 3)	2 (1, 3)	2 (1, 3)
Number of negative words endorsed, Valid n (% of total)h	945 (98.4%)	947 (99.2%)	159 (99.4%)
Mean (standard deviation)	1.2 (1.1)	0.7 (0.8)	0.8 (0.9)
Median (25th, 75th percentile)	1 (0, 2)	0 (0, 1)	1 (0, 1)

^aThe JHS survey did not ask about race because all members of the JHS are African-American.

^bIncludes mixed, Native American, and Asian-American.

^cIncludes retired, disabled, unemployed, in school, and homemaker.

^dScale range 0–7.

^eScale range 1–6.

^fScale range 1–5.

^gScale range 0–4. Respondents who did not check any of the 12 positive, negative or mixed words were considered to have skipped this item and were excluded.

^hScale range 0–5. Respondents who did not check any of the 12 positive, negative or mixed words were considered to have skipped this item and were excluded.

Across the cohorts, participants’ scores on the genetic testing attitude scale, averaging the three ordinal-scale questions (bad thing/good thing; harmful/beneficial; not important/important), were highly positive. JHS, FHS and FHS-Omni respondents had median scores of 5.0, 4.7 and 4.3 respectively. On the genetics attitude measure, JHS, FHS and FHS-Omni respondents endorsed a median of 2 of the 4 positive words, whereas they endorsed a median of 1, 0 and 1 of the 5 negative words respectively. The most commonly endorsed positive words were “hopeful” (60%−70%) and “optimistic” (44%−64%). The most commonly endorsed negative words were “cautious” (35%−38%) and “concerned” (25%−55%). Fewer than 15% of respondents in all cohorts were “worried,” and 5% or fewer were “pessimistic” or “horrified” (Figure 1).

Figure 1:

Percentages of total respondents who endorsed each word, by cohort. Black bars represent positive words, hashed bars represent ambivalent words, and white bars represent negative words.

Factors associated with more negative attitudes

Final logistic regression models are presented in Table 2. For JHS, greater genetics knowledge was associated with lower odds of having more negative attitudes (OR=0.87 per one-point increase in knowledge, 95% CI 0.80–0.96, p=0.004). For FHS, personal experience with genetic testing was associated with lower odds of more negative attitudes (OR=0.45, 95% CI 0.23–0.90, p=0.023). In the FHS-Omni cohort, there were no significant associations between respondent characteristics and more negative attitudes.

Table 2.

Participant characteristics associated with more negative attitudes towards genetic testing, endorsing more positive words, and endorsing more negative words.

Endpoint and Associated Characteristics	JHS		FHS		FHS Omni
	OR (95% CI)a	p-value	OR (95% CI)	p-value	OR (95% CI)	p-value
More negative attitudes towards genetic testingb,c
Genetics knowledge	0.87 (0.80–0.96)	0.004			---
Experience with genetic testing			0.45 (0.23–0.90)	0.023	---
Endorsing more positive wordsd
Genetics knowledge	1.29 (1.16–1.43)	<0.001
Subjective numeracy	1.28 (1.13–1.45)	<0.001	1.17 (1.04–1.31)	0.008
Not employedf					2.15 (1.04–4.74)	0.04
Endorsing more negative wordsc,e
Subjective numeracy	0.88 (0.80–0.98)	0.017			---
Religious Attendanceg			1.20 (1.04–1.38)	0.012	---

^aOR, odds ratio. 95% CI, 95% confidence interval. For dichotomized variables (experience with genetic testing, work status, and frequent religious attendance), odds ratios represent the adjusted relative odds of the outcome variable associated with having the stated characteristic. For index variables (genetics knowledge and subjective numeracy), odds ratios represent the adjusted relative odds of the outcome variable associated with a 1-point increase in the scale.

^bAttitude scale dichotomized at bottom 26^th percentile vs. top 74^th percentile. JHS Model Chi2 8.48, p=0.004, n=882; FHS Model Chi2 6.23, p=0.013, n=911.

^cThere were no significant associations in the FHS-Omni cohort.

^dPositive word score dichotomized at top 33^rd percentile vs. bottom 67^th percentile. JHS Model Chi2 57.34, p<0.001, n=877; FHS Model Chi2 7.27, p=0.007, n=929; FHS Omni. Model Chi2 4.40, p=0.036, n=143.

^eNegative word score dichotomized at top 26^th percentile vs. bottom 74^th percentile. JHS Model Chi2 5.67, p=0.017, n=913; FHS Model Chi2 6.22, p=0.013, n=873.

^fIncludes retired, disabled, unemployed, in school, and homemaker.

^gReligious attendance treated as a continuous variable; there was no significant difference in models that treated religious attendance as a categorical variable vs. a continuous variable.

Factors associated with endorsing more positive words

For JHS, genetics knowledge (OR=1.29 per one-point increase in knowledge, 95% CI 1.16–1.43, p<0.001) and higher subjective numeracy (OR=1.28 per one-point increase in numeracy, 95% CI 1.13–1.45, p<0.001) were associated with higher odds of endorsing more positive words. For FHS, higher subjective numeracy was associated with higher odds of endorsing more positive words (OR=1.17 per one-point increase in numeracy, 95% CI 1.04–1.31, p=0.008). For FHS-Omni, respondents who were not employed had higher odds of endorsing more positive words (OR=2.15, 95% CI 1.04–4.47, p=0.040).

Factors associated with endorsing more negative words

For JHS, higher subjective numeracy was associated with lower odds of endorsing more negative words (OR=0.88 per one-point increase in numeracy, 95% CI 0.80–0.98, p=0.017). For FHS, respondents who more frequently attended religious services had higher odds of endorsing more negative words (OR=1.20, 95% CI 1.04–1.38, p=0.012). For the FHS-Omni cohort, there were no significant associations between respondent characteristics and endorsing negative words.

DISCUSSION

Participants in the JHS, FHS and FHS-Omni cohorts have generally positive attitudes towards genetics and genetic testing. They have high median scores on the three-question genetic testing attitude scale. When asked to select words reflecting their attitudes towards genetics from a list, they endorse more positive words than negative words. These observations are consistent with our expectations for members of highly engaged research cohorts who have consented to participate in the genetics research components of their studies.

Supporting our hypotheses, higher subjective numeracy, greater knowledge of genetics and genetic testing, and personal experience with genetic testing were associated with more positive attitudes. These findings are reassuring—it would be concerning if participants who were most knowledgeable or experienced were most likely to hold negative attitudes, or conversely if participants who were least knowledgeable and experienced were most likely to hold positive attitudes. Adjusting for numeracy, genetic knowledge and other covariates, we did not find an association between attitude and higher educational attainment, which might be due to complex attitudes towards different genetic testing contexts, as seen in prior studies.

We found that frequent attendance at religious services was associated with more negative attitudes towards genetics in the FHS cohort, and that not being employed was associated with more positive attitudes in the Omni cohort. Religious attendance among JHS participants was overwhelmingly high, perhaps explaining why the effect was only seen in FHS. Although we did not have an a priori hypothesis related to employment status, being retired might eliminate concerns about employment discrimination, allowing people to focus more on the potential benefits of genetics (Waltz, Cadigan, Prince, Skinner, & Henderson, 2018).

Our study did not distinguish between attitudes towards genetic testing and attitudes towards genomic sequencing because participants were unlikely to draw those distinctions in the context of their parent studies. The terms “genetic technology” and “genome research” are both used in the parent studies’ educational materials referring to genomics studies. Both JHS and FHS started genome-scale analyses in 2005–6, well before this survey was conducted, and the studies’ newsletters and educational materials kept pace with ongoing research. The consent documents, however, use the term “genetic studies” and “genetic data” even after the studies had adopted genomic methods. The current survey also used the more familiar “genetics” terminology. While participants were therefore educated about the expansion from candidate gene testing to genomic analyses, it is unlikely that these distinctions are relevant to their general attitudes towards genetics. Whether or not participants distinguish between genomics and genetics, it is possible that they would have greater concerns about non-targeted analyses. Additional research on nuanced attitudes towards targeted as compared with non-targeted analyses within research cohorts would be appropriate.

The observation that there is a range of attitudes towards genetics even among people who have enrolled in studies and agreed to genetic testing is among the most important findings of our research. Because all respondents had already consented to genetics research, the findings in this study are primarily relevant to satisfaction and research engagement among enrolled participants rather than to willingness to participate in genetics research in the first place. Investigators should recognize that attitudes towards genetics are not likely to be uniformly positive even among consenting enrollees, and they should invest in ongoing participant education about genetics and transparently communicate their oversight practices to promote participant trust and confidence. This observation and these recommendations have particular relevance for studies such as All of Us that depend on the trust, engagement, and ongoing commitment of large numbers of participants across numerous sites and over an extended period of time.

Consideration of whether these results are generalizable to a less engaged cohort is important. JHS and FHS have both invested in building lasting relationships between researchers and the communities. Participants feel a sense of pride in and identify with being part of these cohorts, sometimes over the course of generations. A multi-site, national cohort with various modes of enrollment may not be able to achieve similarly high engagement, with potential implications for attitudes towards genetic analysis and genetics research. JHS and FHS do an exemplary job of educating and informing participants about genetic analysis and related research activities through regular newsletters and through the informed consent process at each clinic visit. Participants have greater access to information about genetic analysis than similar people who are not research participants. There was, nevertheless, a wide distribution of knowledge scores within each cohort, enabling us to identify relationships between knowledge and attitudes.

Although we observed some differences across cohorts, the general picture that emerged was consistent. In general, greater experience with and knowledge of genetics was associated with more positive attitudes. Individually, each of the three cohorts had very little racial or ethnic variation, but together, they offered geographic and racial diversity. Because our findings were fundamentally consistent across three distinct cohorts, they are likely generalizable to more diverse cohorts such as All of Us.

Limitations

Several limitations of this study are worth noting. First, the data were collected in 2011–12, so questions could arise about the current relevance of the findings. As noted in the introduction, general attitudes towards genetics have become more positive over time. However, there is no reason to expect changes in the relationships between individual characteristics and attitudes towards genetics, which are the primary findings of our study. In addition, the relationships we found are similar to findings about attitudes from other studies in clinical or research contexts described in the introduction, so they are likely to be consistent associations over time. A second limitation is that respondents had already consented to the genetic testing component of their cohort studies; thus, findings from this sample may not generalize to individuals who have declined, or have not yet agreed, to genetic testing in the context of research. Reassuringly, however, very few JHS participants and virtually no FHS participants had declined participation in the genetic testing components of their respective studies. The participants in this survey were therefore likely representative of the full cohorts from which they were drawn, and the findings are likely generalizable to participants in other large, highly engaged cohort studies. Third, the FHS-Omni cohort is underpowered to detect small effect sizes, so it is possible that additional associations would have been found if the sample had been large enough. Fourth, although response rates were high in all cohorts, and we did not observe substantial demographic differences between survey respondents and non-respondents, it is possible that non-respondents would have different attitudes towards genetics and genetic testing. Fifth, the attitude measures used in this study did not specifically differentiate between respondents’ specific views about genetic testing in the context of research studies compared to clinical care, a distinction that future research should address. Finally, future studies would benefit from the inclusion of more contemporary genetic knowledge scales that have been developed since the survey was fielded (Kaphingst et al., 2012; Langer et al., 2017).

BEST PRACTICES

Knowing more about the factors associated with attitudes about genetics and genetic testing can help policymakers and researchers (1) design research that minimizes risks and protects people from potential harms, (2) target outreach and education to help potential participants thoughtfully evaluate genetics research, and (3) develop strategies for communicating with participants once they have joined a study to maximize trust, engagement, retention, and satisfaction.

Researchers have an ethical obligation to promote realistic expectations of risks and potential benefits among prospective and enrolled research participants, which entails proactively combating both excessive enthusiasm and excessive skepticism. Our findings suggest that combating excessive skepticism even after enrollment may be more critical because participants who were less knowledgeable and less experienced were more skeptical. Our findings further suggest that participant education may be effective in alleviating concerns about the risks of genetic testing and genomic research. Researchers should be aware that less knowledgeable participants may have more negative attitudes and therefore may be more likely to withdraw from research, potentially biasing the population of enrolled participants and compromising the generalizability of findings. In order to achieve the goals of fair subject selection and to maximize generalizability, study coordinators should specifically target information delivery to potential participants who have less knowledge and experience.

RESEARCH AGENDA

This study surveyed participants regarding their general attitudes towards genetics and genetic testing. Further work, especially qualitative interviews or focus groups, is needed to identify the specific positive and negative expectations that participants have about genetics, including any misconceptions that require clarification in the context of informed consent and participant education. Furthermore, the implications of attitudes towards genetics for long-term satisfaction with participation should be studied directly.

Short-term educational interventions could enhance genetic literacy and numeracy, thereby improving the quality of informed consent for and engagement with research involving genetic analyses. Future directions for this work would be to develop educational interventions and test whether they are effective at shifting attitudes towards genetics.

EDUCATIONAL IMPLICATIONS

Genetics researchers and study team members need to be aware of the role of attitudes towards genetics in recruitment and retention of research participants. Ongoing education and engagement with existing research participants that lasts beyond initial recruitment and informed consent processes is essential.

Conclusion

Within groups of highly committed research participants, greater experience with and knowledge of genetics are associated with more positive attitudes towards genetics and genetic testing. We expect that positive attitudes will be associated in turn with greater engagement and satisfaction with research. It is possible that targeting outreach to people more likely to have negative attitudes towards genetics would indeed increase their willingness to stay committed to long-term research studies that include genetic components. Short-term interventions, such as giving examples of common genetic testing scenarios and drawing people’s attention to their prior experience with genetic testing, could enhance comfort with genetic analysis and thereby increase participant satisfaction and engagement.