Noninvasive prenatal testing (NIPT) is now widely offered to test for fetal aneuploidies,1,2 and although professional societies have differing recommendations about when to use NIPT as a first-line screening test,3,4 clinical uptake is increasing rapidly.5 The next step in prenatal testing will be to use cell-free fetal DNA from maternal blood6 to sequence the entire fetal genome. Noninvasive prenatal whole genome sequencing will significantly increase access to fetal genetic information, raising ethical concerns regarding which categories of information should or should not be offered, how these results will be returned, and what parents will do with the results.7 Although prenatal whole genome sequencing is not yet clinically available, its feasibility has been demonstrated.8,9 Many factors, such as prenatal whole genome sequencing's clinical and commercial value, the decreasing costs of sequencing, and historical trends demonstrating significant consumer demand for more genetic information, suggest that prenatal whole genome sequencing will likely become integrated into prenatal care.6,10,11
With access to a plethora of scientifically valid fetal genetic information, parents might have to make difficult decisions about whether to undergo prenatal whole genome sequencing and which results to learn.12 Genetic counselors can provide some guidance,13 but there are an estimated 3.95 million births14 in the United States annually, and approximately 4,200 licensed genetic counselors—a shortage relative to both current and anticipated demand.15,16 The role of providing guidance will likely fall on obstetricians and other clinicians, which is concerning in light of a 2016 survey showing that many American obstetrician–gynecologists (ob-gyns) have received little to no genetics training or preparation for interpreting and returning genetic test results.17 Research is needed on how parents will make decisions about what prenatal whole genome sequencing results to receive and what forms of guidance they expect from clinicians.
We created a quantitative survey built on a previous survey of obstetricians' views on the ethics of prenatal whole genome sequencing.17 Preadministration cognitive testing was performed to ensure optimal understanding of the survey, and adjustments were made to clarify question order and to reconcile survey language with how women discuss and understand the use of prenatal testing. Pilot testing was conducted with ten patients and clinicians in the Inova healthcare system, located in northern Virginia. In response to the collected comments, the survey was edited to improve understanding of and comfort with the questions and to optimize the overall structure and flow of the instrument.
The primary endpoint for the survey was the average preference score for seeking prenatal whole genome sequencing information. A sample size calculation was conducted based on a two-sided confidence interval for one proportion with assumed point estimate equal to 50% (most conservative) and a desired confidence interval width of ±4%. At the 90% confidence level, the resulting sample size was 446, and at the 95% confidence level, it was 623.
The survey was administered to women who were 8 or more weeks pregnant and receiving prenatal care at three Inova clinics. Inova Translational Medicine Institute research coordinators approached patients as they waited for scheduled appointments. Participants were offered English or Spanish versions of the survey and could take the survey on paper or electronically through Survey Monkey. The research protocol was approved by the National Human Genome Research Institute Institutional Review Board and the requirement to obtain written documentation of informed consent was waived.
As prenatal whole genome sequencing is not clinically available, the survey (Appendix 1, available online at http://links.lww.com/AOG/B281) included an introductory page to educate participants about prenatal whole genome sequencing and how it differs from current NIPT.
At the start of the survey, participants were asked to envision undergoing prenatal whole genome sequencing and were presented with eight categories of genomic results: one nonmedical and seven medical genomic categories organized by age of disease onset, disease severity, prevalence, and actionability (ie, availability of a clinical intervention or preventative action that can treat or modify the course of disease, Table 1). The order of the categories was randomized to avoid an ordering effect. Respondents were asked to indicate their level of interest in receiving different categories of results and then selected the primary reason why they would choose or not choose to receive the results from a standardized list of reasons.
The second half of the survey explored a range of topics, including the appropriate role of clinicians in prenatal whole genome sequencing decision-making, desirability of state or federal regulation of prenatal whole genome sequencing, and comfort with uncertain genomic information. The survey concluded with demographic questions and a standardized genetic literacy measure, the Genetic Literacy and Comprehension test.18
For each of the eight categories of genetic information in the first portion of the survey, we condensed the responses into yes (“Definitely” or “Probably”), or no (“Definitely not” or “Probably not”) and calculated the percentage of respondents who wanted and did not want to receive the information. Responses for all pairs of genomic information categories were compared using McNemar's test for 2×2 tables.
We assigned numeric values to responses: definitely would not want the information (−2), probably would not want (−1), probably would want 1), definitely would want 2). We then averaged each participant's responses across all eight categories to create a single measure of information-seeking tendencies. A linear model was fitted to evaluate which factors predicted information-seeking tendencies, which was analyzed as a numeric variable. Eleven candidate explanatory variables were considered for the model of information-seeking preferences: genomic literacy, religiosity, response to the statement “the more information parents have about their baby's genes the better,” abortion, age, race–ethnicity, education, previous experience with a genetic disorder, income, comfort with uncertainty, and marital status. All predictors, except the first, are categorical variables, some nominal and some ordinal. The “best” set of predictors was selected based on Schwarz Bayesian Information Criterion.19
A second model was generated to evaluate which factors predicted how much help respondents would want from their doctors when making decisions about prenatal whole genome sequencing. An ordinal logistic regression was used. Fourteen predictors were considered for the model of help from clinicians: age, race–ethnicity, education level, income, marital status, genomic literacy score, religiosity, response to the statements “the more information parents have about their baby's genes the better,” “parents should be able to access all medically relevant genetic information that they want to know,” “parents should be able to access all nonmedical genetic information that they want to know,” “it is appropriate for physicians to provide their opinion about the kinds of genetic information that parents should learn about their babies,” “the state or federal government should decide what categories of fetal genetic information can and cannot be returned,” and response to the questions “are there any circumstances in which you would choose to end a pregnancy,” and “do you want the results for these genes.”
The “best” model was determined based on a combination of approaches: stepwise regression, elimination of one predictor at a time, P values of estimated parameters, R-square, percent concordance, and percent discordance.
Inova Translational Medicine Institute research coordinators invited 805 people to take the survey between June and August 2017; 605 agreed, and 553 returned a survey that was at least partially completed, resulting in a response rate of 68.7%. The vast majority of respondents completed all questions (65.3%) or skipped only one (23.3%) or two (7.2%); missing data were distributed evenly across all questions. Respondents reported a range of income and education levels, varying degrees of religiosity, and diverse racial backgrounds. The majority of respondents were 21–40 years old (Table 2).
Of the categories of genomic information presented (Table 1), women were least likely to want to learn about nonmedical traits (40.0%). Women were most likely to want information regarding serious treatable childhood-onset conditions (89.7%). For other categories, the proportion of women who wanted to receive the information ranged from 65.5% to 83.4% (Appendix 2, available online at http://links.lww.com/AOG/B301).
All pairwise comparisons between categories of genetic information were statistically significantly different from one another (P<.05), with the exception of fatal adult-onset and common untreatable conditions (P=.054), common treatable conditions and developmental delays–learning disabilities (P=.257), developmental delays–learning disabilities and fatal congenital conditions (P=.834), and common treatable and fatal congenital conditions (P=.248) (Appendix 3, available online at http://links.lww.com/AOG/B301). Of the women who wanted information about developmental delays–learning disabilities and common treatable conditions, the most reported reason was to prepare financially, medically, or psychologically for a child with the condition. However, for fatal congenital conditions, although the highest percentage of women (35.4%) also wanted the information to prepare for their child, a similar percentage (31%) wanted it to consider ending the pregnancy.
The first model was created to test which variables predicted average information-seeking tendencies. It produced three statistically significant results: 1) participant agreement with the statement, “the more information parents have about their baby's genes, the better” (P<.001); 2) participant willingness to receive uncertain genetic information (P<.001); and 3) race–ethnicity (P<.001) (Fig. 1A).
Because the first two variables measure a quality analogous to our information-seeking tendencies score, we removed them as possible candidate variables. The subsequent model produced two significant predictors: genetic literacy (P=.042, estimate=0.048, 95% CI [0.002–0.096]) and views on abortion (P<.001) (Fig. 1B).
Generally, most women wanted genetic information that pertained to future medical conditions (eg, childhood cancers, heart disease). Women most frequently cited wanting medical categories of information in order “to prepare financially, medically, or psychologically for a child with special needs.” This held true with fatal congenital conditions, where 35.4% of women wanted the information to prepare, but, unlike other conditions, 31% of women wanted the information “to inform a decision about whether to end the pregnancy”—a close second.
For women who did not want to receive genetic information about medical categories, the most frequently cited reasons were “it is not relevant for my reproductive decision-making since I would not consider ending a pregnancy” and “I would worry that the information would increase my stress or the stress of my family” (Fig. 2).
For nonmedical information, such as eye color, height, or athletic ability, the reasons most often selected by respondents were “just to know it, but would not act on the information” and “to prepare financially, medically, or psychologically for a child with special needs” (Fig. 2). Frequently selected reasons among women not interested in nonmedical information included: “this kind of information is not important to me” (26.2%) and “it is not relevant for my reproductive decision-making since I would not consider ending a pregnancy” (15.0%) (Fig. 2, Appendix 4, available online at http://links.lww.com/AOG/B301).
Respondents were asked how much they agreed with the statement that “it is appropriate for physicians to provide their opinion about the kinds of genomic information that parents should learn about their babies” and 70.2% either somewhat or strongly agreed (Appendix 5, available online at http://links.lww.com/AOG/B301, Appendix 6 available online at http://links.lww.com/AOG/B301). In contrast, when asked how much they agreed with “the state or federal government should decide what categories of fetal genetic information can and cannot be returned” 14.8% either somewhat or strongly agreed (Appendixes 5 and, http://links.lww.com/AOG/B301).
We also asked what kind of help respondents would want from their doctor when deciding which categories of information to receive from prenatal whole genome sequencing. 10.5% wanted clear recommendations, 44.7% wanted clear recommendations plus all options presented, 26.2% wanted all options presented with a joint decision-making process, and 13.2% wanted all options presented but to make an independent decision (Fig. 3).
The model for determining what kind of assistance patients wanted from clinicians identified three statistically relevant predictor variables: genetic literacy (P=.046), education (P=.036), and views on abortion (P=.058) (Appendix 7, available online at http://links.lww.com/AOG/B301). The initial model included the question “it is appropriate for physicians to provide their opinion about the kinds of genetic information that parents should learn about their babies” which was excluded from final model owing to overlap with the response variable.
Based on the ordinal logistic regression model controlling for genetic literacy score, education level and views on abortion, respondents with a high school degree or less tended to prefer clear recommendations from their doctor relative to the other educational groups; and those with some college education tended to prefer to make an independent decision relative to the other groups. Women with lower genetic literacy preferred clear recommendations, although it was somewhat difficult to characterize their preferences owing to small sample sizes among lower scores: 70.9% of respondents scored a 6 or above. Women who said there were no circumstances under which they would end a pregnancy and women who would consider pregnancy termination both had stronger preferences towards clear recommendations than respondents who would consider ending a pregnancy only in very limited circumstances.
Women were asked how much they wanted scientifically uncertain genetic information related to autism on a scale of 1 (definitely would want to know) to 5 (definitely would not want to know). The majority of women (55.5%) answered 1 or 2, indicating they would want this uncertain information (Appendix 8, available online at http://links.lww.com/AOG/B301).
Based on existing prenatal testing literature, we hypothesized that respondents would generally want genetic information about their fetuses but that women would most likely request information on severe, actionable, and early-onset conditions—the types of conditions for which we currently test. Some results supported this hypothesis: the smallest proportion of women wanted to receive nonmedical information, and nearly all respondents wanted information about serious treatable childhood-onset conditions. Even so, more than half of respondents chose to receive all medical categories, without further differentiation based on actionability, severity, prevalence, and age of onset. There were significant differences between many pairs of categories, but other, dissimilar categories did not show differences suggesting that as prenatal whole genome sequencing is integrated into prenatal care, clinicians should not assume that the same reasons underlie decisions to receive different types of information.
Higher genetic literacy predicted increased inclination towards information seeking. Respondents who would not consider abortion were less likely to seek out genetic information about their fetuses than those who would, which was an intuitive finding. Race–ethnicity was also a significant predictor, with those who identified as Hispanic or Latina choosing to receive less genetic information than those in other racial or ethnic groups.
Our model is not causal, however, and many demographic factors correlate with race and ethnicity, including income and education. The confluence of factors makes it difficult to disentangle race and ethnicity from income level, particularly among women who identified as Hispanic or Latina. Furthermore, although some subscribe to the idea that more information is generally better, we ascribe no moral valence to the phrase “information-seeking,” and made efforts to frame our survey questions in as neutral a manner as possible. We recognize that there are a variety of individual-specific reasons a woman may or may not want to receive prenatal whole genome sequencing results. Clinicians should be mindful that the set of information women will want to receive is not one-size-fits-all.
Generally, women who wanted a particular category of medical genetic information about their children wanted it to prepare emotionally, psychologically, and financially. The exception was nonmedical information, where the most popular reason selected was “just to know it, but [I] would not act on the information.” These data partially alleviate concerns that the medical importance or certainty of prenatal whole genome sequencing results will be overstated and lead many women to terminate pregnancies or make misinformed decisions about their future fertility.7
The two main reasons that women did not want to learn certain kinds of genetic information were 1) a worry that the information would increase stress and 2) they would not consider having an abortion. One exception was nonmedical information, where most women simply found the information unimportant. These reasons align with the results of a previous survey, which found that ob-gyns were primarily worried that prenatal whole genome sequencing results would cause increased parental anxiety.17 Future research may help identify the specific reasons women would find prenatal whole genome sequencing results stressful.20 It is also notable that women who did not want information often saw abortion as the most salient action they could take, whereas women who wanted the information did not frequently appeal to pregnancy termination as their primary motivation for seeking the information.
Prenatal whole genome sequencing offers significantly more fetal information than women can currently receive, and it is not surprising that when faced with a tremendous range of information, many women want recommendations from their clinicians. Our data suggest that most women prefer a directive interaction with their clinician when deciding what types of genetic information to receive from prenatal whole genome sequencing. This represents a departure from the current state of genetic counseling, where nondirectiveness is a central tenet,21 and is contrary to the 45.3% of ob-gyns who said in a previous survey that they should not be at all directive when counseling patients on prenatal whole genome sequencing.17 Given this clear patient desire for guidance, there is a vital opportunity for the American College of Obstetricians and Gynecologists to provide leadership and recommendations as prenatal whole genome sequencing is adopted into clinical practice.
Our survey has a number of potential limitations. First, we conducted our survey in northern Virginia, and there are regional differences and demographic factors we did not assess that may influence preferences. Second, when describing categories of information available through prenatal whole genome sequencing, we included example conditions. The conditions were meant to help participants envision what types of disorders fit into each category, but respondents could have fixated on the examples rather than generalizing the point. Third, because prenatal whole genome sequencing is not yet clinically available, we were asking women to make hypothetical decisions, which might differ from the actual decisions they would make in a clinical setting. Fourth, in asking women why they wanted (or did not want) information, we required them to pick the single most important reason from a preselected list. Many reasons may be important to women and our survey may not fully capture them all. Social desirability bias, which posits that people are reluctant to report actions or preferences they perceive as less socially acceptable,22 may also have led to underreporting of socially controversial reasons including abortion. Finally, although our response rate was reasonable, there could have been an undetected systematic difference between people who enrolled and those who did not, introducing a nonresponse bias.
Although this study is an important step towards understanding the perspectives of pregnant women on prenatal whole genome sequencing, there is much future research to be done. Other questions that remain to be answered include how women would make decisions about prenatal whole genome sequencing in collaboration with their partners, what additional reasons women have for wanting to receive or not receive certain information, how women would really act after learning the results of prenatal whole genome sequencing, and what ethical concerns women have about prenatal whole genome sequencing. Future work could also assess women's knowledge of genomics in greater depth and longitudinally, as well as their understanding of current prenatal genetic testing options, to create decision-making tools.
In the coming years, professional societies such as the American College of Obstetricians and Gynecologists and the American College of Medical Genetics and Genomics will likely draft guidance on expanded prenatal genetic testing, including prenatal whole genome sequencing. They must pay particular attention to the views of pregnant women and women of reproductive age, who are the central stakeholders. Furthermore, if obstetricians and other nongenetic counselor clinicians are to become the primary gatekeepers of prenatal whole genome sequencing, they will need additional training on how to present and return fetal genetic information in a manner that corresponds closely to the needs and preferences of their patients.
1. Minear MA, Alessi S, Allyse M, Michie M, Chandrasekharan S. Noninvasive prenatal genetic testing: current and emerging ethical, legal, and social issues. Annu Rev Genomics Hum Genet 2015;16:369–98.
2. Allyse MA, Wick MJ. Noninvasive prenatal genetic screening using cell-free DNA. JAMA 2018;320:591–2.
3. Gregg AR, Skotko BG, Benkendorf JL, Monaghan KG, Bajaj K, Best RG, et al. Noninvasive prenatal screening for fetal aneuploidy, 2016 update: a position statement of the American College of Medical Genetics and Genomics. Genet Med 2016;18:1056–65.
4. Screening for fetal aneuploidy. Practice Bulletin No. 163. American College of Obstetricians and Gynecologists. Obstet Gynecol 2016;127:e123–37.
5. Green ED, Rubin EM, Olson MV. The future of DNA sequencing. Nature 2017;550:179–81.
6. Vermeesch JR, Voet T, Devriendt K. Prenatal and pre-implantation genetic diagnosis. Nat Rev Genet 2016;17:643–56.
7. Donley G, Hull SC, Berkman BE. Prenatal whole genome sequencing: just because we can, should we? Hastings Cent Rep 2012;42:28–40.
8. Kitzman JO, Snyder MW, Ventura M, Lewis AP, Qiu R, Simmons LE, et al. Noninvasive whole-genome sequencing of a human fetus. Sci Transl Med 2012;4:137ra176.
9. International Society for Prenatal Diagnosis, Society for Maternal-Fetal Medicine, Perinatal Quality Foundation. Joint position statement from the International Society for Prenatal Diagnosis (ISPD), the Society for Maternal-Fetal Medicine (SMFM), and the Perinatal Quality Foundation (PQF) on the use of genome-wide sequencing for fetal diagnosis. Prenat Diagn 2018;38:6–9.
10. Regalado A. 10 breakthrough technologies: prenatal DNA sequencing. MIT Technology Review; May 2013. Available at: https://www.technologyreview.com/s/513691/prenatal-dna-sequencing/
. Retrieved January 17, 2019.
11. Hayden EC. Technology: the $1,000 genome. Nature 2014;507:294–5.
12. Wou K, Weitz T, McCormack C, Wynn J, Spiegel E, Giordano J, et al. Parental perceptions of prenatal whole exome sequencing (PPPWES) study. Prenat Diagn 2018;38:801–11.
13. Allyse MA, Sayres LC, Havard M, King JS, Greely HT, Hudgins L, et al. Best ethical practices for clinicians and laboratories in the provision of noninvasive prenatal testing. Prenat Diagn 2013;33:656–61.
14. Martin JA, Hamilton BE, Osterman MJK, Driscoll AK, Drake P. Births: final data for 2016. Natl Vital Stat Rep 2018;67:1–55.
15. American Board of Genetic Counseling, Inc. Find a certified genetic counselor. Available at: https://www.abgc.net/about-genetic-counseling/find-a-certified-counselor/
. Retrieved July 18, 2018.
16. Hoskovec JM, Bennett RL, Carey ME, DaVanzo JE, Dougherty M, Hahn SE, et al. Projecting the supply and demand for certified genetic counselors: a workforce study. J Genet Couns 2018;27:16–20.
17. Bayefsky MJ, White A, Wakim P, Hull SC, Wasserman D, Chen S, et al. Views of American OB/GYNs on the ethics of prenatal whole-genome sequencing. Prenat Diagn 2016;36:1250–6.
18. Hooker GW, Peay H, Erby L, Bayless T, Biesecker BB, Roter DL. Genetic literacy and patient perceptions of IBD testing utility and disease control: a randomized vignette study of genetic testing. Inflamm Bowel Dis 2014;20:901–8.
19. Beal DJ. Information criteria methods in SAS for multiple linear regression models. Available at: https://analytics.ncsu.edu/sesug/2007/SA05.pdf
. Retrieved January 16, 2019.
20. Genetti CA, Schwartz TS, Robinson JO, VanNoy GE, Petersen D, Pereira S, et al. Parental interest in genomic sequencing of newborns: enrollment experience from the BabySeq project. Genet Med 2018 [Epub ahead of print].
21. Veach PM, Bartels DM, Leroy BS. Coming full circle: a reciprocal-engagement model of genetic counseling practice. J Genet Couns 2007;16:713–28.
22. Krumpal I. Determinants of social desirability bias in sensitive surveys: a literature review. Qual Quantity 2011;47:2025–47.