You are here

Article Text

Article menu
Clinical guidelines
Original article
Impact of direct-to-consumer genomic testing at long term follow-up
  1. Cinnamon S Bloss1,
  2. Nathan E Wineinger1,
  3. Burcu F Darst1,
  4. Nicholas J Schork1,2,
  5. Eric J Topol1,2,3
  1. 1Scripps Genomic Medicine, Scripps Translational Science Institute, and Scripps Health, La Jolla, California, USA
  2. 2Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla, California, USA
  3. 3Scripps Clinic Medical Group, La Jolla, California, USA
  1. Correspondence to Dr Cinnamon S Bloss, Scripps Genomic Medicine, Scripps Translational Science Institute, 3344 N. Torrey Pines Court, Suite 300, La Jolla, CA 92037, USA; cbloss{at}scripps.edu

Abstract

Background There are few empirical data to inform the debate surrounding the use and regulation of direct-to-consumer (DTC) genome-wide disease risk tests. This study aimed to determine the long term psychological, behavioural, and clinical impacts of genomic risk testing for common disease.

Methods The Scripps Genomic Health Initiative is a prospective longitudinal cohort study of adults who purchased the Navigenics Health Compass, a commercially available genomic test. Web based assessments were administered at baseline, short (3 months), and long term (1 year) follow-up.

Results 2240 participants completed either or both follow-ups and a subset of 1325 completed long term follow-up. There were no significant differences from baseline in anxiety (p=0.50), fat intake (p=0.34), or exercise (p=0.39) at long term follow-up, and 96.8% of the sample had no test related distress. Longitudinal linear mixed model analyses were consistent with results of cross-sectional analyses. Screening test completion was associated with sharing genomic test results with a physician (36.0% shared; p<0.001) and perceived utility of the test (61.5% high perceived utility; p=0.002), but was not associated with the genomic risk estimate values themselves.

Conclusions Over a third of DTC genomic test recipients shared their results with their own physician during an approximate 1 year follow-up period, and this sharing was associated with higher screening test completion. Genomic testing was not associated with long term psychological risks, and most participants reportedly perceived the test to be of high personal utility.

  • direct-to-consumer
  • genetic testing
  • personalized medicine
  • genomic risk testing
  • consumer genomics

https://doi.org/10.1136/jmedgenet-2012-101207

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    Direct-to-consumer (DTC) genome-wide disease risk tests continue to be sold over the internet, a practice which remains controversial.1 Common criticisms of the tests include a lack of demonstrated clinical validity and utility, as well as a lack of appropriate involvement of healthcare providers or regulatory oversight guidance. DTC genomic testing companies themselves have been accused of deceptive marketing practices2 and the inability to ensure the privacy of consumers’ genomic data.3 Other concerns include the potential for the tests to cause psychological or other harms to consumers or to inappropriately burden the healthcare system. Although empirical data are beginning to emerge that can inform aspects of the debate surrounding these tests and their availability without a healthcare intermediary, there are still many unanswered questions.

    The Scripps Genomic Health Initiative (SGHI) is a longitudinal cohort study designed to assess the psychological, behavioural, and clinical impacts of DTC genomic risk testing for common disease.4 SGHI participants purchased a commercially available genomic test at a subsidised rate between 2008 and 2009 and were administered baseline, short term, and long term follow-up web based health assessments that included standardised measures of anxiety, diet, exercise, test related distress, and health screening behaviours. From this effort we recently reported one of the first large scale studies on consumer response to genomic risk testing and found no measureable impacts on any of the outcomes, which were assessed at short term follow-up (average 6 months).5 An obvious limitation of this initial study, however, was the short follow-up interval observed. Here we report data from our long term follow-up assessment, which was collected just over 1 year after participants received their genomic test results.

    Materials and methods

    The SGHI was approved by the Scripps Office for the Protection of Research Subjects and Institutional Review Boards. Informed consent was obtained electronically from each participant. The details of our methods have been published previously4 ,5 and will only be briefly summarised; additional information is included in the online supplementary material.

    Study sample and design

    The SGHI was initiated in October 2008. Participants were adults recruited primarily from health and technology companies, and procedures pertaining to enrolment have been previously described.4 ,5 Participants were administered web based health assessments at baseline, as well as 3 months (referred to as ‘short term’ follow-up) and 1 year (referred to as ‘long term’ follow-up) after receiving their genomic test results. The primary analyses presented here are based on data from the long term follow-up. Secondary analyses include linear mixed effects model analyses that leverage data from both time points.

    Procedures for administration of the 1 year follow-up were such that once 365 days had elapsed from the time the participant received their genomic testing results, an email was sent to the participant with a link to the web based survey utility asking them to complete the assessment. Importantly, participants who were considered lost to short term follow-up were still contacted and asked to complete the long term assessment. A total of three email requests were sent within a 6 week time frame after which point, if the individual still had not completed the long term follow-up assessment, they were considered lost to long term follow-up.

    Intervention

    We examined the impact of DTC genomic risk testing with the Navigenics Health Compass,6 at that time, a commercially available genomic test, which was developed based on previously reported genetic risk loci disease associations (see online supplementary figure S1). Information regarding the test and its content as initially released to SGHI participants can be found in a previous publication5 and in the online supplementary material. During the long term follow-up period of the SGHI, Navigenics developed and eventually released risk results for five new conditions, including deep vein thrombosis, melanoma, sarcoidosis, haemochromatosis, and lactose intolerance. In this report, we have included risk estimates for these new conditions in analyses evaluating psychological, behavioural, and clinical screening response to DTC genomic risk testing.

    As part of the study, participants agreed to provide us with access to their genomic risk reports. This enabled us to assess the overall impact of genomic testing, as well as the impact of the personalised risk estimates disclosed. Study participants were also offered, at no charge, genetic counselling provided by Navigenics’ staff of board certified genetic counsellors. In addition, as previously described, Navigenics conducted proactive outreach to subgroups of their customers (which included study participants) based on the customer's genomic risk results.

    Instruments

    Baseline assessment included items to assess demographics, family health-span history, personal/individual health-span history, and attitudes about genetic testing. Both the baseline and follow-up assessments additionally included measures of anxiety, fat intake, and exercise behaviour to enable us to assess change in these areas. Situational anxiety was assessed with the state anxiety subscale of the Spielberger State-Trait Anxiety Inventory (STAI),7 fat intake with the Block Fat Screener,8 and exercise behaviour with the Godin Leisure-Time Exercise Questionnaire (GLTEQ).9 ,10 Both short and long term follow-up assessments included evaluation of test related distress using the Impact of Events Scale-Revised (IES-R).11 ,12 In addition, 13 health screening behaviours were assessed by asking about actual completion of a given screening test, as well as self-reported intention to complete the screening test with greater frequency since undergoing genome-wide disease risk testing. These 13 tests included the following: thyroid test, skin exam, ophthalmic exam, glucose exam, electrocardiogram, colonoscopy, cholesterol test, chest x-ray, cardiac stress test, blood test, self-breast exam, mammogram, and prostate specific antigen test.

    Participants were also asked at both follow-ups whether they had shared their results with their own physician or healthcare provider and/or whether they had spoken with a Navigenics genetic counsellor about their results. Finally, unique to the 1 year assessment were items to gauge literacy (ie, ‘Do you understand your test results?’), perceived consistency with family health history (ie, ‘Do you feel your genetic test results are generally consistent with your previous health history and your family health history?’), and perceived utility of the genomic test (ie, ‘In general, do you feel your genetic test results are useful to you?’).

    Outcome measures

    Primary outcomes were changes from baseline in subjects’ anxiety symptoms, fat intake, and exercise behaviour, as well as levels of test related distress and self-reported completion and intention to complete health screening behaviours at long term follow-up.

    Statistical analysis

    All statistical analyses were conducted using the statistical software packages SPSS 14.0, R 2.13.2,13 and VasserStats web utility for computing z-tests and confidence intervals for proportions.14 Two-sided t tests, Mann-Whitney U tests or χ2 tests were used to compare baseline variables between individuals who completed follow-up versus those who were lost to follow-up (table 1). Repeated measures analysis of variance and the within subjects effect of time controlling for eight covariates (age, sex, education, ethnicity (Caucasian yes/no), income, health related occupation, long term follow-up interval in days, and completion or non-completion of short term follow-up) was used to assess the extent to which baseline and long term follow-up scores on the STAI, Fat Screener, and GLTEQ differed. A similar analysis was done comparing IES-R scores, total screening tests completed, and total screening tests with intended increased frequency between short and long term follow-up. Percentages were also used to qualitatively describe scores on the IES-R at long term follow-up.

    View this table:
    Table 1

    Baseline comparisons between individuals who completed follow-up versus those who were lost to follow-up

    Associations between our outcomes and the genetic risk estimates disclosed were evaluated using linear regression controlling for the eight covariates described above and baseline scores on each measure. Specifically, follow-up STAI, Fat Screener, and GLTEQ scores were regressed on (1) average estimated lifetime risk (ELTR) across all of the conditions for which genetic results were viewed, (2) proportion of conditions colour coded orange indicating increased genetic risk (again, of those viewed), and (3) ELTR and colour coded risk for each of the 28 condition specific risk estimates. See a previous publication for a description of the composite genetic risk variables constructed for analysis (ie, items 1 and 2 above).5 This was similarly done for IES-R scores (without a baseline score). Each screening test was matched with its corresponding condition(s) and tested using logistic regression with the screening test (either actual or intended completion) as the dependent variable and our eight covariates plus the condition specific risk estimate as the independent variable.

    We also assessed the proportion of participants who reported discussing their results with a Navigenics genetic counsellor or their own physician or healthcare provider, as well as the extent to which these factors were associated with study outcomes. Association between condition specific genomic risk estimates, sharing results with a physician, and their interaction was evaluated using a logistic regression model with the completion of a corresponding screening test as the dependent variable. We performed similar analyses for self-reported perceptions of understanding, utility, and consistency of genomic results.

    Finally, longitudinal analyses were conducted to evaluate relationships with the condition specific risk estimates and our outcomes of interest across both short and long term follow-ups. Because of the longitudinal nature of the study and the fact that the data consist of non-uniform numbers of repeated measurements at non-uniform follow-up intervals, linear mixed-effects model analysis was used in which a random effect associated with the individual study participants was assumed. Each of the above statistical models included our eight covariates, as well as the baseline score (when applicable).

    Results

    Study sample and demographics

    Figure 1 depicts enrolment numbers and outcomes. Descriptive statistics for demographic and outcome variables are shown in table 1, as are baseline comparisons of these variables between individuals who completed long term follow-up (n=1325) versus those who were lost to long term follow-up, as well as those who completed any follow-up (n=2240) versus those who were lost to any follow-up. The same baseline comparisons were also performed between individuals who completed both follow-ups (n=1122) versus only the long term follow-up (n=203) (see online supplementary table S1). Composite and condition specific genetic risk estimates were also compared between the same groups (see online supplementary tables S2–S7).

    Figure 1
    Figure 1

    Enrolment, study groups, and outcomes. Individuals who withdrew cited financial reasons, insurance concerns, and change of mind. Non-compliance was when an individual had not completed the baseline health assessment after three email requests. Protocol deviation indicates that the individual's genetic results were unintentionally released to them before completion of the baseline health assessment. Technical issues means that the individual submitted duplicate or triplicate surveys. QC, quality control.

    The number of individuals who completed any follow-up is consistent with an overall response rate of 62%, which is relatively high for internet based survey research.15 The average long term follow-up interval was 14 months (SD=1.3).

    Long term response to genome-wide disease risk testing

    Psychological

    There were no significant differences in state anxiety between baseline and long term follow-up (table 2). At long term follow-up, 96.8% of individuals had a total score of ≤8 on the IES-R Avoidance and Intrusion subscales indicating no test related distress, and nearly all of the individuals in the sample (99.7%) had a total score of ≤23 indicating no clinically significant test related distress. Test related distress significantly decreased between short and long term follow-up (table 2).

    View this table:
    Table 2

    Primary outcome measures at baseline and follow-up in 1325 participants

    Diet and exercise

    We found no significant differences in fat intake or exercise behaviour between baseline and long term follow-up (table 2).

    Health screening behaviours

    There were no significant differences in the total number of screening tests completed between short and long term follow-up (table 2). Although there were also no significant differences between the number of screening tests participants intended to complete with greater frequency at long term relative to short term follow-up, 42.4% of the sample continued to report at long term follow-up that there was at least one or more screening tests they intended to complete with greater frequency post-genomic testing.

    We also evaluated the extent to which intended screening at short term follow-up predicted actual screening at long term follow-up. Although the total number of screening tests participants intended to complete with greater frequency at short term follow-up was a significant predictor of the total number of screening tests actually completed at long term follow-up, the effect size was small (β=.059, p=.02).

    Study outcomes and genomic risk estimates

    There were no significant associations between composite measures of genetic risk and long term follow-up state anxiety, test related distress, fat intake, exercise behaviour scores, or total number of screening tests completed. Alternatively, a significant association was observed between composite measures of risk and total number of screening tests participants intended to complete with greater frequency (average orange: β=0.059, p=0.03).

    Associations between long term follow-up scores on our behavioural measures and condition specific risk estimates are shown in online supplementary tables S8–S11. Associations between condition specific risk estimates and completion of specific screening behaviours at long term follow-up, as well as the intention to complete specific screening behaviours with increased frequency post-testing, are shown in online supplementary tables S12 and S13. For the sample as a whole, few, even nominally significant, associations were observed between genetic risk and psychological functioning (see online supplementary figure S2), behavioural outcomes, or completion of specific screening tests at long term follow-up.

    Longitudinal analyses incorporating short and long term follow-up data

    Results of linear mixed model analysis evaluating relationships with the condition specific risk estimates and anxiety, test related distress, fat intake, and exercise behaviour across both short and long term follow-ups are shown in online supplementary tables S9 and S11. Few significant associations are observed, and the results are consistent with cross-sectional analyses. Associations between condition specific risk estimates and actual and intended completion of specific screening tests at the most recent follow-up available for the full cohort are shown in online supplementary table S13. Again, findings are generally consistent with cross-sectional analyses.

    Sharing results with a physician or genetic counsellor

    In the long term follow-up sample, 14.1% of individuals reported discussing their results with a Navigenics genetic counsellor, and 39.5% reported sharing their results with their own physician or healthcare provider (see online supplementary tables S14 and table S15 for percentages across the entire sample). Across both follow-ups, speaking with a Navigenics genetic counsellor was not associated with fat intake, exercise behaviour, or the intention to complete screening tests with greater frequency, but was associated with anxiety, test related distress, and actual screening test completion. Spontaneously sharing results with a physician or healthcare provider was not associated with anxiety, test related distress, or fat intake, but was associated with higher exercise, actual screening test completion, and intention to increase frequency of screening (see online supplementary table S15).

    Given the high percentage of the sample that spontaneously shared genomic results with a personal physician and the association between sharing and increased screening test completion, a behaviour that is consistent with increased health care resource utilisation, we further investigated the relationship between physician sharing and screening. Specifically, we assessed the extent to which sharing was associated with completion of individual screening tests, the pattern of results within different age groups (as a way to evaluate whether the increased screening was appropriate), as well as the extent to which the relationship between sharing and screening test completion was moderated by the condition specific genomic risk estimates disclosed.

    Physician sharing was significantly associated with completion of the majority of individual screening tests, and this finding was observed at both short and long term follow-up (figure 2A, see online supplementary tables S16–S18). In terms of the pattern of results of this main effect for different age groups, the largest number of tests with statistically significant differences between physician sharers and non-sharers was seen for 30–39-year-olds, 40–49-year-olds, and 50–59-year-olds. The youngest (<30 years old) and oldest age groups (60+ years old) showed the fewest number of tests that differed in terms of rates of completion (see online supplementary table S19, figure 3). This pattern of results was not observed for intention to increase screening (see online supplementary figure S3). In terms of the role of genetic risk with respect to actual screening test completion, for six of the screening test/condition pairs evaluated, the condition specific genomic risk estimate moderated the effect of sharing on screening such that individuals who both shared and who received a high genomic risk estimate were more likely to screen (figure 2C; see online supplementary figure S4).

    Figure 2
    Figure 2

    Forrest plots depicting the effect of physician sharing, genetic risk, and the interaction between these two variables on screening test completion at follow-up. For each screening test/condition pair, logistic regression was performed separately for short and long term follow-up data in which completion of the screening test (yes/no) was the dependent variable and the independent variables were eight covariates, physician sharing (yes/no), condition specific genetic risk (high/low), and the interaction between physician sharing and genetic risk. For each screening test/condition pair, the top line indicates the OR and 95% CI at short term follow-up and the bottom line reflects the same data at long term follow-up. The first panel reflects the main effect of physician sharing (A), the second, the main effect of genetic risk (B), and the third the effect of the interaction (C). Red lines indicate statistical significance. The actual ORs and p values for all the comparisons are shown in online supplementary table S16. The tests and condition pairs are as follows: thyroid test/Graves disease, skin exam/melanoma, skin exam/psoriasis, ophthalmic exam/glaucoma, ophthalmic exam/macular degeneration, glucose exam/type 2 diabetes, electrocardiogram/atrial fibrillation, electrocardiogram/heart attack, colonoscopy/colon cancer, colonoscopy/Crohn's disease, cholesterol level/heart attack, cholesterol level/type 2 diabetes, chest x-ray/lung cancer, chest x-ray/sarcoidosis, cardiac stress test/atrial fibrillation, cardiac stress test/heart attack, blood test/coeliac disease, blood test/colon cancer, blood test/Crohn's disease, blood test/haemochromatosis, blood test/lactose intolerance, blood test/lupus, blood test/rheumatoid arthritis, self-breast exam/breast cancer, mammogram/breast cancer, and prostate specific antigen/prostate cancer.

    Figure 3
    Figure 3

    Total number of screening tests completed and summed, shown as a function of participant age group and whether or not results were shared with a physician. Error bars reflect the 95% CI for the mean and the number of tests is based on data from the most recent follow-up available for 2240 participants. Total number completed at follow-up is higher among older individuals and physician sharers. The youngest and oldest groups showed the fewest number of tests that differed between sharers and non-sharers.

    We performed similar interaction analyses of genomic risk with respect to age and family history of disease (see online supplementary tables S20 and S21). Although the main effect of age was almost uniformly associated with individual screening test completion (consistent with the age bin analyses presented above), and family history showed variable main effect associations, the genomic risk estimate values were not found to moderate any of these relationships.

    Perceived understanding, consistency and utility of results

    Analyses of participants’ responses to items gauging understanding, perceived consistency, and perceived utility of their genomic risk results are presented in online supplementary figure S5. The majority of individuals indicated that they felt they did understand their results (74.5%), that their results were either somewhat (34.1%) or very useful (61.5%), and that their results were either somewhat (44.7%) or very (34.6%) consistent with their prior health history. Associations between these variables and study outcomes are depicted in online supplementary tables S22 and S23. Individuals who reported that they understood their results showed lower anxiety and test related distress at long term follow-up. Perceiving the genomic test results to be useful was associated with both total number of screening tests completed and intention to screen, and reporting that results were consistent with prior health history was associated with intention to screen.

    Discussion

    Prospective assessment of a selected group of individuals who underwent DTC genomic risk testing for common disease and completed long term follow-up just over 1 year after receiving results showed no measureable changes in anxiety level, dietary fat, or exercise behaviour. In addition, more than 96% of the sample had no test related distress. Over a third of genomic test recipients spontaneously shared their results with their own physician or healthcare provider, and this sharing was associated with higher screening test completion. For a subset of the individual screening tests, there was a significant interaction between sharing and genomic risk. That is, individuals who both shared and who received an estimate of high genomic risk were more likely to complete screening. High perceived personal utility of the genomic test was also associated with screening test completion.

    We were surprised to find that in our full cohort of 2240 individuals, 36% (n=807) reported that they shared their results with their own physician or healthcare provider at either or both follow-up time points. This rate of sharing is higher than that found in at least one other study,16 and one explanation for this may be related to the co-pay charged for the genomic test in our study, which may have motivated participants to share with a physician as a way of ‘getting their money's worth’ so-to-speak. Much continues to be made of the lack of necessary training and knowledge in genomics, as a group, among physicians.17 This, taken together with our finding of high rates of sharing, suggests an ongoing void and possible barrier to the advancement of genomic medicine centred on physician genomic education. Along these same lines, we found that the sharing of genomic test results with a physician—a response to genomic testing itself—was one of the only factors we measured that was associated with actual behaviour change post-genomic testing. Specifically, even after controlling for important covariates, physician sharing was associated with higher health screening test completion (non-sharers=4.5 total screening tests completed versus sharers=6.4 total screening tests completed) at long term follow-up. We also note, however, that this finding may not be generalisable beyond the confines of the US healthcare environment, particularly to countries where access to physicians is free at the point of delivery.

    One often raised concern about DTC genomic testing is that it will lead to increases in unnecessary health screening tests and thus inappropriate economic burdens on the healthcare system.18 To explore this, we examined total screening test completion, as well as patterns of specific test completion as a function of age (figure 3 and see online supplementary table S19, respectively). The youngest age group (<30 years old) in our sample (ie, those individuals least likely to ‘need’ screening according to most current guidelines for most screening tests) showed the fewest number of tests that were completed more frequently among physician sharers versus non-sharers. Specifically, two screening tests were completed by young physician sharers with greater frequency relative to non-sharers, including cholesterol and blood test. We note that the National Cholesterol Education Program’s Adult Treatment Panel III (ATP III) recommends a fasting lipoprotein profile (total cholesterol, low density lipoprotein cholesterol, high density lipoprotein cholesterol, and triglycerides) in all adults over the age of 20 once every 5 years.19 While our data cannot speak to this issue directly or definitively, these analyses are suggestive of the notion that at least some fraction of any increases in screening motivated by genomic testing (or more specifically, sharing genomic test results with a physician) may be consistent with motivating compliance with current recommended guidelines.18 ,20

    A primary limitation of this study is that we evaluated a sample of convenience, and our longitudinal cohort design did not include the use of a control group. Also, although our sample has been shown to be representative of Navigenics customers and the current broader population of consumers of genome-wide risk tests,5 given that it is quite homogenous (eg, highly educated, high income), our conclusions may not be applicable to other groups of individuals should the market expand, or to the general population. Further, the study encountered differential dropout between baseline and follow-up. Although all measured variables that were statistically different between individuals who completed follow-up versus were lost to follow-up were included as covariates in the analysis, it is possible that this may have resulted in some degree of bias. Finally, the SGHI relies on brief, web based, self-report assessment of health behaviours, which can be less reliable than in-person assessment and less sensitive than more objective measures.21 Major strengths of our study include the large sample size studied, the use of validated measures of psychological functioning and behaviour, the multiple follow-ups observed, and the evaluation of a previously commercially available test.

    Across the cohort as a whole, we failed to find any evidence of long term psychological impacts of testing, including adverse psychological outcomes. In fact, test related distress significantly decreased at long term follow-up relative to short term follow-up. We likewise did not observe significant improvements in lifestyle behavioural factors,22 including diet and exercise, at long term follow-up. Screening test completion also did not significantly increase across the sample as a whole and we failed to find an association between screening and the actual genetic disease risk estimates disclosed. Rather, screening test completion was correlated to a large extent with physician sharing, as well as the general perception among participants that the genomic test was useful (see online supplementary table S22).

    It is noteworthy that a very large fraction of our long term follow-up sample also reported perceiving that they generally understood their test results and that the test had high personal utility. This finding suggests that DTC genome-wide disease risk profiling is considered to have high utility by a reasonable fraction of consumers. This, coupled with the lack of adverse psychological outcomes observed, is consistent with ongoing direct access by patients to genomic testing, assuming appropriate regulatory oversight is in place.

    We note that the American Medical Association has released a policy recommendation that genetic testing be carried out under the personal supervision of a qualified health care professional.23 At odds with this, however, is the poor knowledge base of physicians in the area of genomics,17 in addition to the very low numbers of genetic counsellors and medical geneticists in the USA. Specifically, in the USA and Canada there are currently only 3026 certified genetic counsellors, including one in the state of Wyoming and none in West Virginia.24 This is compared to the 850 085 physicians with active licences to practise in the USA in 2010.25 These statistics leave patients with limited options. Furthermore, there is evidence that some groups of consumers may be hesitant to share or discuss genomic or genetic testing with a healthcare provider due to privacy concerns.26 ,27 These circumstances, together with our findings that highlight consumers’ spontaneous sharing of genomic information with their physician, suggest the possibility that a new model in which patients and physicians are partners—that is, patient direct access with support from physicians as desired—may be optimal. To mandate physicians or health providers as gatekeepers to genomic testing would at this point seem inconsistent with emerging empirical data showing a lack of adverse response to (and reasonable levels of interest in) genomic testing among consumers.

    Acknowledgments

    We acknowledge the support Laura Ornowski, MS, of Scripps who assisted with data collection, as well as Vance Vanier, MD, Michele Cargill, PhD, and Elana Silver, MS, of Navigenics who, along with their genetic counsellors and other staff, helped support the project.

    References

    1. ↵
      1. Bloss CS,
      2. Darst BF,
      3. Topol EJ,
      4. Schork NJ
      . Direct-to-consumer personalized genomic testing. Hum Mol Genet 2011;20(R2):R132–41.
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. Trent RJ
      . Pathology practice and pharmacogenomics. Pharmacogenomics 2010;11:105–11.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Caulfield T,
      2. McGuire AL
      . Direct-to-consumer genetic testing: perceptions, problems, and policy responses. Annu Rev Med 2012;63:23–33.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Bloss CS,
      2. Ornowski L,
      3. Silver E,
      4. Cargill M,
      5. Vanier V,
      6. Schork NJ,
      7. Topol EJ
      . Consumer perceptions of direct-to-consumer personalized genomic risk assessments. Genet Med 2010;12:556–66.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Bloss CS,
      2. Schork NJ,
      3. Topol EJ
      . Effect of direct-to-consumer genomewide profiling to assess disease risk. N Engl J Med 2011;364:524–34.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Borry P,
      2. Howard HC,
      3. Senecal K,
      4. Avard D
      . Health-related direct-to-consumer genetic testing: a review of companies’ policies with regard to genetic testing in minors. Fam Cancer 2010;9:51–9.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Spielberger CD,
      2. Gorsuch RL,
      3. Lushene PR,
      4. Vagg PR,
      5. Jacobs AG
      . Manual for the state-trait anxiety inventory (Form Y). Palo Alto: Consulting Psychologists Press, Incorporated, 1983.
    8. ↵
      1. Block G,
      2. Gillespie C,
      3. Rosenbaum EH,
      4. Jenson C
      . A rapid food screener to assess fat and fruit and vegetable intake. Am J Prev Med 2000;18:284–8.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Godin G,
      2. Shephard RJ
      . A simple method to assess exercise behavior in the community. Can J Appl Sport Sci 1985;10:141–6.
      OpenUrlPubMed
    10. ↵
      1. Jacobs DR Jr.,
      2. Ainsworth BE,
      3. Hartman TJ,
      4. Leon AS
      . A simultaneous evaluation of 10 commonly used physical activity questionnaires. Med Sci Sports Exerc 1993;25:81–91.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Weiss DS,
      2. Marmar CR
      . The impact of event scale—revised. In: Wilson JP, Keane TM Assessing psychological trauma and PTSD. New York: Guilford Press, 1997:399–411.
    12. ↵
      1. Creamer M,
      2. Bell R,
      3. Failla S
      . Psychometric properties of the Impact of Event Scale—Revised. Behav Res Ther 2003;41:1489–96.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      R: A language and environment for statistical computing. Secondary R: A language and environment for statistical computing 2008. http://www.R-project.org.
    14. ↵
      1. Grosse SD,
      2. Kalman L,
      3. Khoury MJ
      . Evaluation of the validity and utility of genetic testing for rare diseases. Adv Exp Med Biol 2010;686:115–31.
      OpenUrlCrossRefPubMed
    15. ↵
      1. Yetter G,
      2. Capaccioli K
      . Differences in responses to web and paper surveys among school professionals. Behav Res Methods 42:266–72.
    16. ↵
      1. Kaphingst KA,
      2. McBride CM,
      3. Wade C,
      4. Alford SH,
      5. Reid R,
      6. Larson E,
      7. Baxevanis AD,
      8. Brody LC
      . Patients’ understanding of and responses to multiplex genetic susceptibility test results. Genet Med 2012. Published Online First: Epub Date. doi: 10.1038/gim.2012.22 gim201222 [pii]
    17. ↵
      1. Healy M
      . As genetic testing races ahead, doctors are left behind. Los Angeles Times 2009.
    18. ↵
      1. McGuire AL,
      2. Burke W
      . An unwelcome side effect of direct-to-consumer personal genome testing: raiding the medical commons. JAMA 2008;300:2669–71.
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      Executive summary of the third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA 2001;285:2486–97.
      OpenUrlCrossRefPubMedWeb of Science
    20. ↵
      1. Ng PC,
      2. Murray SS,
      3. Levy S,
      4. Venter JC
      . An agenda for personalized medicine. Nature 2009;461:724–6.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Reid RJ,
      2. McBride CM,
      3. Alford SH,
      4. Price C,
      5. Baxevanis AD,
      6. Brody LC,
      7. Larson EB
      . Association between health-service use and multiplex genetic testing. Genet Med 2012. Published Online First: Epub Date. doi:10.1038/gim.2012.52
    22. ↵
      1. Bloss CS,
      2. Madlensky L,
      3. Schork NJ,
      4. Topol EJ
      . Genomic information as a behavioral health intervention: can it work? Per Med 2011;8:659–67.
      OpenUrlCrossRefPubMed
    23. ↵
      AMA. D-480.987 Direct-to-Consumer Marketing and Availability of Genetic Testing. Secondary D-480.987 Direct-to-Consumer Marketing and Availability of Genetic Testing. http://www.ama-assn.org/ama1/pub/upload/mm/31/policyd-480–987.pdf
    24. ↵
      1. O'Daniel JM,
      2. Haga SB,
      3. Willard HF
      . Considerations for the impact of personal genome information: a study of genomic profiling among genetics and genomics professionals. J Genet Couns 2010;19:387–401.
      OpenUrlCrossRefPubMedWeb of Science
    25. ↵
      1. Young A,
      2. Chaudhry HJ,
      3. Rhyne J,
      4. Dugan M
      . A census of actively licensed physicians in the United States, 2010. J Med Regul 2011;96:10–20.
      OpenUrl
    26. ↵
      1. Darst BF,
      2. Madlensky L,
      3. Schork NJ,
      4. Topol EJ,
      5. Bloss CS
      . Characterization of genomic test consumers who spontaneously share results with their physician. Health Commun in press.
    27. ↵
      1. Roberts ME,
      2. Riegert-Johnson DL,
      3. Thomas BC
      . Self diagnosis of Lynch syndrome using direct to consumer genetic testing: a case study. J Genet Couns 2011;20:327–9.
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Contributors CSB contributed to the design of the study, performed statistical analyses, drafted the manuscript, and edited/revised the manuscript for critical scientific content. NEW and BFD performed statistical analyses, drafted parts of the manuscript, and edited the manuscript for critical scientific content. NJS and EJT contributed to the design of the study and edited the manuscript for critical scientific content. All authors have read and approved the final manuscript.

    • Funding This work was supported in part by a NIH/NHGRI R21 grant (1R21HG005747-01; PI: CSB), a NIH/NCRR flagship Clinical and Translational Science Award grant (1UL1RR025774-01; PI: EJT), and Scripps Genomic Medicine Division of Scripps Health.

    • Competing interests None.

    • Ethics approval Scripps Health Institutional Review Board.

    • Provenance and peer review Not commissioned; externally peer reviewed.