Direct‐to‐consumer genetic testing for factor V Leiden and prothrombin 20210G>A: the consumer experience

Abstract Background Clinical genetic testing for inherited predisposition to venous thromboembolism (VTE) is common among patients and their families. However, there is incomplete consensus about which individuals should receive testing, and the relative risks and benefits. Methods We assessed outcomes of receiving direct‐to‐consumer (DTC) results for the two most common genetic risk factors for VTE, factor V Leiden in the F5 gene (FVL) and prothrombin 20210G>A in the F2 gene (PT). Two thousand three hundred fifty‐four customers (1244 variant‐positive and 1110 variant‐negative individuals) of the personal genetics company 23andMe, Inc., who had received results online for F5 and F2 variants, participated in an online survey‐based study. Participants responded to questions about perception of VTE risk, discussion of results with healthcare providers (HCPs) and recommendations received, actions taken to control risk, emotional responses to receiving risk results, and perceived value of the information. Results Most participants (90% of variant‐positive individuals, 99% of variant‐negative individuals) had not previously been tested for F5 and/or F2 variants. The majority of variant‐positive individuals correctly perceived that they were at higher than average risk for developing VTE. These individuals reported moderate rates of discussing results with HCPs (41%); receiving prevention advice from HCPs (31%), and making behavioral changes to control risk (e.g., exercising more, 30%). A minority (36%) of variant‐positive individuals worried more after receiving VTE results. Nevertheless, most participants reported that knowing their risk had been an advantage (78% variant‐positive and 58% variant‐negative) and were satisfied knowing their genetic probability for VTE (81% variant‐positive and 67% variant‐negative). Conclusion Consumers reported moderate rates of behavioral change and perceived personal benefit from receiving DTC genetic results for VTE risk.


| INTRODUCTION
Venous thromboembolism (VTE) is characterized by deep vein thrombosis (DVT), pulmonary embolism (PE), or both. VTE affects 300,000-600,000 individuals in the United States each year, can cause long-term complications and is associated with high morbidity and mortality (Beckman, Hooper, Critchley, & Ortel, 2010;Behravesh et al., 2017). Genetic, acquired, and situational factors influence risk for VTE.
Risk is compounded by numerous acquired and situational factors, including oral contraceptive use, hormone replacement therapy (HRT), pregnancy, and prolonged periods of inactivity, among others (Kujovich, 1993b(Kujovich, , 1993cVarga & Kujovich, 2012). Management of risk in asymptomatic individuals may include avoiding compounding risk factors and/or temporary prophylaxis. For instance, known thrombophilia risk may influence decisions about contraception and HRT and may prompt increased vigilance and/ or prophylaxis during pregnancy and postpartum (American College of Obstetricians & Gynecologists Women's Health Care Physicians, 2013; Bates et al., 2012;Kujovich, 1993c;Trenor et al., 2011).
There is incomplete agreement about which individuals should be tested for these two variants to inform treatment and prevention decisions (De Stefano & Rossi, 2013;Zhang et al., 2018). Clinical guidelines such as those of the Evaluation of Genomic Applications in Practice and Prevention Working Group recommend testing in circumstances where test results are likely to influence clinical management and discourage routine testing for patients with VTE and their family members (Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group, 2011;Grody, Griffin, Taylor, Korf, & Heit, 2001;Kujovich, 1993c;Stevens et al., 2016). Other guidelines diverge, and in practice, clinical genetic testing in these patients and their families is widespread (Stevens et al., 2016;Varga & Kujovich, 2012). There have been public health efforts to raise awareness of VTE, including, in 2008, the U.S. Surgeon General's recommendation to raise consumer awareness about inherited thrombophilia (Beckman et al., 2010; Office of the Surgeon General (US) & National Heart, Lung, and Blood Institute (US), 2008), and there are indications of consumer interest in understanding inherited risk (Hellmann, Leslie, & Moll, 2003).
Given the apparent gap between clinical guidelines and consumer and public health organizations' interest in greater awareness, it is useful to assess the impact of direct-to-consumer (DTC) testing for thrombophilia. For over 10 years 23andMe, Inc., a personal genetics company, has offered DTC results for numerous genetic risk factors, including select variants associated with thrombophilia (Carere et al., 2014;Francke et al., 2013).
Here, we report data from a nine-month study on consumer response to DTC genetic testing for FVL and PT. We discuss recall of test results, sharing of test results with health care providers (HCPs) and family members, HCP recommendations, behavioral changes based on test results, and emotional responses to testing.

| Ethical compliance
All participants in the study provided consent and answered surveys online according to 23andMe's research protocol, which was approved by Ethical & Independent Review Services, an AAHRPP-accredited institutional review board. Participants whose free-text responses are quoted gave 23andMe explicit permission to do so. The study conforms to U.S. Federal Policy for the Protection of Human Subjects.

| Recruitment
Study participants were customers of 23andMe who obtained 23andMe's personal genomic services, including a VTE genetic risk report, between November 2007 and November 2013 (Carere et al., 2014). An example of the VTE report offered during this time is in Figure S1.
23andMe research participants who were over the age of 30, had logged into their 23andMe.com account within the two-year period prior to November 2013, were not part of | 3 of 13 ELSON Et aL.
any other 23andMe disease research study, and had opted to receive health results were invited to participate. Eight thousand five hundred thirty-six variant-positive individuals with one or more FVL or PT variant and 11,353 variant-negative individuals were invited to participate.

| Data collection
Study participation consisted of responding to at least one of two online surveys. Survey 1 was offered to participants from March to April 2015 and included up to 102 questions, depending on survey branching logic. Survey 1 addressed participants' recall of the genetic test result and subsequent perception of VTE risk; personal and family history of VTE, including prior testing for FVL and PT; discussing results with HCPs and family; behavioral changes in response to test results; emotional responses to testing; and demographics. Survey 2 was offered nine months later, from December 2015 to January 2016, to participants who had taken Survey 1 and consisted of up to 51 questions. Survey 2 further interrogated participants' health outcomes, attitudes, and understanding of the causes and consequences of VTE. Both surveys asked whether participants had viewed their 23andMe VTE report-Survey 1, if they had ever viewed the report; Survey 2, if they had viewed it within the previous nine months. Survey questions are provided in Supplemental Material.
One thousand two hundred forty-four variant-positive and 1110 variant-negative individuals consented and responded to Survey 1. Seven hundred fifty-one cases and 574 controls took both surveys (Table 1). Survey response rates, stratified by sex and variant status, are provided in Table 1 and Table  S1. Rates of response were roughly consistent with previously reported rates (Tung et al., 2011).

| Data analysis
The data were analyzed by variant-positive versus variantnegative status and sometimes stratified by genotype and sex. Table 1, Tables S1-S3 include data from all participants who provided the relevant responses. Tables 2-8 and  Table S3 include data only from participants who reported having viewed their 23andMe VTE report prior to study participation; these participants were asked to not refer to their 23andMe reports when responding. Tables 1, 3-5, 7A, 8, and Tables S1-S3 include data from Survey 1; Tables 1, 2, 6, 7B, and Table S1 include data from Survey 2. Free text responses were read and coded by two authors (SLE and HMR). Recurring themes identified in at least five responses were reported; other themes were added to the "Other" category (Table 8).
Fifty-four percent of study participants were female. Participants were primarily of European ancestry and had an average age of 57; most were at least college-educated and were relatively high-income earners (Table 1).
Eighty percent of cases and 46% of controls reported having viewed their 23andMe VTE report before taking Survey 1. A minority of participants (10% of cases and 1% of controls) knew their genetic risk for VTE prior to receiving their 23andMe genetic test results (Table 1). Among those who had had prior testing, it was often obtained because of a personal or family history of VTE (Table S2).
The surveys included questions about personal and family history of VTE. Approximately 9% of cases and 3% of controls who took Survey 1 reported a personal history of VTE. Twenty-three percent of cases and 13% of controls reported having a first-degree biological relative who had been diagnosed with VTE.

| Test results and risk perception
Participants who had viewed their 23andMe VTE report prior to study participation responded to questions about their 23andMe results and VTE risk. Depending on genotype, 83%-90% of cases correctly reported that their genetic risk for VTE was higher than average, and 74%-86% of cases correctly responded that they had at least one risk variant (Table 2). In contrast, 47% of controls correctly responded that their risk was average or below average (34% were unsure), and 26% of controls correctly responded that they did not have a risk variant (59% were unsure). (Of note, the VTE report relevant to this study indicated slightly higher than average VTE risk for individuals with non-O blood types. Forty-five of the 51 controls who reported higher risk for VTE, and 36 of the 42 controls who responded that they had one or more mutations, had received a report that indicated slightly elevated risk due to non-O blood type). Sixty-nine percent of cases with two risk variants (FVL or PT homozygotes, FVL/PT double heterozygotes) correctly responded that their lifetime risk for VTE was greater than 40%, and 55% of cases with one risk variant correctly reported that their lifetime risk for VTE was 10%-40% (Table S3). Most cases and controls were unable to recall their specific genotypes.

| Discussing results with family, friends, and HCPs
More cases than controls reported discussing their 23andMe results with their family, friends, and HCPs (Table 3). Sixtyfive percent of cases and 17% of controls discussed their results with a first-degree family member. Sixty percent of cases and 22% of controls, discussed with a spouse or significant other. Forty-one percent of cases and 7% of controls discussed their results with a HCP. Among cases who discussed with a physician, most discussed with a primary care physician (88%), followed by other specialties. Cases discussed their results with fewer non-physician providers than physicians (Table 3).

| HCP recommendations
Thirty-one percent of cases and 14% of controls reported that HCPs recommended changes in lifestyle or medications, or additional testing (Table 4). For cases, the most common recommendations were to wear compression socks (36%), exercise more (34%), and lose weight (28%). Smoking cessation was recommended to 5% of cases and none of the five controls, possibly reflecting a low prevalence of smoking in the study population.
Medication for primary prevention was recommended to 18% of cases, 65% of whom had no history of VTE. Almost all recommendations were for aspirin (data not shown). Anticoagulant (such as heparin or warfarin) use after surgery was recommended to 17% of cases, and 11% of cases were encouraged to use an anticoagulant for a longer duration than originally planned. Fewer than five controls received similar recommendations (data not shown).
Discontinuation of estrogen-containing oral contraceptives or HRT was recommended to, respectively, 15% and 10% of female cases; heparin treatment during pregnancy was recommended to 5% of cases. Zero controls received these recommendations.
Repeat genetic testing in a clinical lab, or testing for other clotting disorders, was recommended to, respectively, 21% and 13% of cases. Testing in relatives was recommended to 15% of cases. Zero controls received this recommendation.

| Behavior and medication changes
Participants answered questions about lifestyle and medication changes prompted by their 23andMe results. Consistent with HCP recommendations, 30% of cases reported beginning to exercise more, 25% took step steps to lose weight, and 12% started to wear compression socks/stockings. In comparison, 12%, 10%, and <2% of controls, respectively, reported making these changes (Table 5). Two percent of cases started taking medication to prevent blood clots, and 4% took a blood thinner such as heparin or warfarin for a longer duration than originally planned. Four and five percent of female cases, respectively, discontinued or changed either an estrogen-containing contraceptive or HRT. Fewer than 1% of controls made any of these medication changes. A lower percentage of cases (43%) than controls (79%) reported no changes in lifestyle or medication. We assessed whether receiving thrombophilia results prompted testing of family members of FVL and PT carriers. A higher percentage of cases than controls reported that their family members had had testing for genetic mutations associated with VTE (Table 6), and more cases than controls (28% versus 3%) reported that their relatives received testing because of their genetic results. Among the minority of cases whose relatives received testing because of their results, 48% reported testing through 23andMe; 34%, through a HCP, and 17% through a combination of a HCP and 23andMe.

| Perceptions of value and emotional responses to testing
Both cases and controls reported benefit in receiving VTE risk results. Eighty-one percent of cases and 67% of controls responded that they were satisfied with knowing their genetic risk for VTE; 1% of cases and controls were unsatisfied, and the rest were neither satisfied nor unsatisfied (Table  7A). Similarly, 78% of cases and 58% of controls reported that knowing their genetic probability for VTE had been an

T A B L E 4 HCP recommendations
When you shared your 23andMe results with your healthcare provider(s), did he/she recommend any changes to your lifestyle, any changes to your medications, or additional testing?

T A B L E 5 Behavioral and medication changes
Have you made any of the following changes as a result of your 23andMe report on genetic risk for VTE? Please check all that apply. advantage, <1% of cases and controls reported that it had been a disadvantage, and the remainder responded "neither an advantage nor disadvantage." Most participants thought that they could "take steps to reduce the probability of VTE" (87% of cases, 85% of controls), and nearly all (97% of cases, 93% of controls) would choose to learn their genetic risk for VTE again if they could do it all over again (Table 7B). Approximately half of participants (52% of cases, 56% of controls) reported that they worried the same after learning their genetic risk for VTE; more cases than controls worried more (36% versus 8%), and more controls than cases worried less (36% versus 12%) (Table 7A).

Cases (n = 911) Controls (n = 457)
Participants had the opportunity to describe in free text the positive and negative impacts of knowing their genetic risk for VTE (Table 8). Among cases, 642 respondents reported positive impacts, and 201 reported negative impacts; 253 and 47 controls, respectively, reported positive and negative impacts. The most common theme for positive impacts for both cases and controls was ability to take preventative measures. Positive impacts described by cases also included the ability to inform HCPs and to recognize symptoms; those described by controls included increased knowledge of VTE or personal risk, and relief with knowing risk status. The most common theme for negative impacts for both cases and controls was actual or potential for worry. Negative impacts described by cases also included unpleasantness of knowing risk for self and/or family, and those described by controls included potential false sense of security.

| DISCUSSION
Our study assessed outcomes of DTC testing for inherited thrombophilia, including recall of test results, actions taken following receipt of test results, sharing with and recommendations from HCPs, emotional responses, and perception of value. Our data contribute to the discussion of controversial issues in the field, related both to testing for inherited thrombophilia and to DTC testing generally. The Evaluation of Genomic Applications in Practice and Prevention Working T A B L E 6 Family testing for thrombophilia

Cases (n = 737) Controls (n = 545)
Child ( Group concluded that there is no evidence that knowledge of F5/F2 status among asymptomatic family members reduced their VTE risk and found that the potential harms may outweigh the benefits (Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group, 2011). Concerns about DTC genetic testing include risk of consumers misinterpreting genetic risk information, potential overutilization of medical resources, and negative psychosocial outcomes (Botkin et al., 2010;Leighton & Valverde, 2011;McGuire & Burke, 2008). However, there are indications of consumer and public health interest in greater understanding of thrombophilia. These is also increasing recognition that individuals may derive personal benefit from receiving genetic health risk information, including the ability to understand inherited risk and make healthy choices to control it ( Grody et al., 2001;Kujovich, 1993bKujovich, , 1993cStevens et al., 2016), we sought to understand if genetic risk information obtained through DTC testing could have personal utility. In this study, recall of genetic test results depended on both primary understanding of the VTE report and ability to remember the report content, in most cases, over one year after receiving results, as participants were instructed to rely on their recollection of the test information to answer survey questions. Among participants who had viewed their results, cases were more likely than controls to correctly respond to the question of whether they had at least one risk variant and to correctly perceive their risk as above average (Table 2). These data suggest that receipt of a positive or impactful result is associated with better comprehension and/or recall of the result. Inability to recall a specific F5 and F2 genotype was common among cases and controls (Table S3). Together these data indicate that participants could retain the underlying message of having a risk variant or being at higher risk based on genetics, even if they did not recall specific details. Questions of whether and how DTC genetic testing alters consumer behavior or health are areas of active investigation (Bloss, Wineinger, Darst, Schork, & Topol, 2013;Carere et al., 2014;Hollands et al., 2016;Kaufman, Bollinger, Dvoskin, & Scott, 2012;van der Wouden et al., 2016). In our study, substantial numbers of cases shared their results with HCPs, family, and friends (Table 3). Over half of cases made behavioral or medication changes, the most common of which were exercising more and taking steps to lose weight (Table  5). These interventions are unlikely to induce harm and may improve health generally. The findings are consistent with prior studies showing that people who consider themselves at higher risk for a disease are more likely to discuss results with HCPs and make behavioral changes (Kaufman et al., 2012), but they are inconsistent with those of a large meta-analysis that found no evidence that communicating DNA-based risk estimates changes behavior (Hollands et al., 2016). Future investigations might explore the nuances of when and how knowledge of genetic risk, obtained either DTC or in the context of medical care, affects behavior and health outcomes.
Despite not being discussed in the 23andMe VTE report, several cases (12%) reported using compression garments for clot prevention. This may indicate that participants received HCP recommendations or obtained health advice through alternative sources. While compression garments are generally safe, there can be contraindications, particularly in the absence of medical support, and there is insufficient evidence to support their use for primary prophylaxis (Lim & Davies, 2014).
Among cases who discussed their results with HCPs, approximately one third received lifestyle, medication, or additional testing recommendations (Table 4). Unexpectedly, we identified a potential trend of HCPs recommending that asymptomatic cases take aspirin for prevention. An outstanding question is whether HCP recommendations were based on thrombophilia risk-which is not part of clinical management guidelines-or other risks. Given the potential for harm with unnecessary aspirin treatment (McNeil et al., 2018), future studies might explore the reasons for and prevalence of HCP recommendations for aspirin, as well as whether these and other recommendations for medication changes are appropriate in light of current management guidelines.
It is noteworthy that about 15% of cases reported that HCPs recommended testing for relatives, even though many clinical guidelines discourage routine testing of carriers' family members. The finding is consistent with evidence that clinical testing of VTE patients and their relatives for thrombophilia is widespread (Stevens et al., 2016;Varga & Kujovich, 2012). Interestingly, while 40% of cases reported that their relatives had received genetic testing for VTEassociated variants, a minority of these relatives (28%) had received testing because of the case's 23andMe results, and approximately, half of these (11% of overall respondents) did so through the medical system (Table 6).
More cases than controls expressed satisfaction with knowing their results and felt that knowing was advantageous (Table 7). Similarly, cases placed more value in knowing their results than controls did. Most participants felt that they could take steps to reduce their risk for VTE, and nearly, all would learn their results again if they could do it over. These results are interesting given that 36% of cases reported worrying more about VTE. Our finding that most participants would choose to know their risk, despite potential for increased worry, is consistent with findings from a previous study that assessed patient experience of a positive FVL test result (Hellmann et al., 2003).
A limitation of this study was that we relied on participants' self-report and lacked full comorbidity profiles, which might have helped explain HCP recommendations and actions taken. Another limitation was the variability in time between participants' receiving the report and responding to the survey, which may have affected accuracy of recollections. Finally, the study population might differ from the general population, including those interested in testing for inherited thrombophilia. The 23andMe participant base is more educated and has higher socioeconomic status than the general U.S. population.

| CONCLUSIONS
This study builds on previous investigations into consumer reactions to DTC testing for disease risk (Carere et al., 2014;Francke et al., 2013), in this instance providing insight into outcomes of testing for VTE risk. We observed moderate rates of cases taking action following receipt of test results (e.g., sharing results with a medical provider or making behavioral changes) and high rates of satisfaction with test results. Further, cases reported positive impacts of receiving test results that included ability to take preventative measures, to inform medical providers, and to recognize signs and symptoms, as well as increased knowledge of VTE and personal risk. Taken together, these results suggest that consumers may experience personal benefit from receiving DTC genetic results for VTE risk that is distinct from the potential benefit in receiving clinically actionable information.