Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Aug 1.
Published in final edited form as: Circ Cardiovasc Genet. 2017 Aug;10(4):e001613. doi: 10.1161/CIRCGENETICS.116.001613

Effect of Disclosing Genetic Risk for Coronary Heart Disease on Information Seeking and Sharing: The MI-GENES Study

Sherry-Ann N Brown 1, Hayan Jouni 1, Tariq S Marroush 1,2, Iftikhar J Kullo 1
PMCID: PMC5547821  NIHMSID: NIHMS889696  PMID: 28779015

Abstract

Background

Whether disclosing genetic risk for coronary heart disease (CHD) to individuals influences information seeking and information sharing is not known. We hypothesized that disclosing genetic risk for coronary heart disease (CHD) to individuals influences information seeking and sharing.

Methods and Results

The Myocardial Infarction Genes (MI-GENES) study randomized participants (n=203) aged 45–65 years who were at intermediate CHD risk based on conventional risk factors and not on statins, to receive their conventional risk score (CRS) alone or also a genetic risk score (GRS) based on 28 variants. CHD risk was disclosed by a genetic counselor and then discussed with a physician. Surveys assessing information seeking were completed before and after risk disclosure. Information sharing was assessed post-disclosure. Six months post-disclosure, GRS participants were more likely than CRS participants to visit a website to learn about CHD (OR 4.88 (1.55–19.13), P=.01), use the internet for information about how genetic factors affect CHD risk (OR 2.11 (1.03–4.47), P=.04), access their CHD risk via a Patient Portal (OR 2.99 (1.35–7.04), P=.01), and discuss their CHD risk with others (OR 3.13 (1.41–7.47), P=.01), particularly their siblings (OR 1.92 (1.06–3.51), P=.03), extended family (OR 3.8 (1.37–12.38), P=.01), co-workers (OR 2.42 (1.09–5.76), P=.03), and primary care provider (PCP) (OR 2.00 (1.08–3.75), P=.03).

Conclusions

Disclosure of a GRS for CHD increased information seeking and sharing.

Clinical Trial Registration

https://clinicaltrials.gov/; Unique Identifier: NCT01936675.

Journals Subject Terms: Clinical Studies, Translational Studies, Cardiovascular Disease, Primary Prevention, Genetics

Keywords: psychology and behavior, genetic testing, genetic polymorphism, coronary heart disease, prevention, information seeking, information sharing, genetic risk score, precision medicine, coronary heart disease risk

Introduction

Enabling and empowering patients to participate in their health care is critical to the implementation of precision (or personalized) medicine and improving health care outcomes 1,2. Increased patient engagement and participation in their care promotes truly personalized healthcare 3,4, and associates with subsequent healthier behaviors, better clinical indicators, and lower health care costs 5. Such patient participation involves an ‘electronic health care journey’ 2, which includes where participants go to seek information about their health and whether they access a limited view of the electronic health record (EHR) in the form of a personal health record (PHR) via the patient portal. The patient journey also involves sharing of health information in the patient’s biological and social networks.

Health information seeking and sharing facilitate informed decision-making, through augmenting the patient’s knowledge of disease processes, preventive behaviors, and therapeutic options, and associate with the practice of healthy behaviors 6,7. The internet enables patients to access health information 8 (from websites or the PHR) and share information about their disease risk with others through email or on social media platforms. Gaining further insight into the patient’s information seeking and sharing behaviors following disclosure of disease risk for health promotion may help ensure available, accessible, and high quality health information, to enable patients (i.e., e-patients 2) to engage with their physicians.

Seeking and sharing of genetic information in particular have implications for the health not only of the index patient, but also of a patient’s family, household members, and friends. 9 Individuals often seek and share their genetic risk information in the midst of their kinship, given that the risk may also be shared by their family members. Not only does this increase awareness in families, but also perception of genetic risk information as a shared threat might catalyze joint interventions among family. 10 This differentiates the impact of disclosing genetic risk information from disclosing non-genetic risk information alone. Nevertheless, for chronic complex diseases (e.g., CHD) with common susceptibility gene variants, risk can be modified by shared environment and behaviors in individuals beyond biological kinship. As a result, sharing risk information with non-biological household members and friends could also help to propagate health information in the patient’s social network. 10

CHD remains the leading cause of morbidity and mortality worldwide, and is consequently a public and global health priority 11. Several prospective cohort or case control studies and a meta-analysis have shown that a multi-locus GRS for CHD can help restratify patients to higher or lower risk for developing CHD 12–19. Studies have suggested that a GRS is predictive of CHD events and clinical benefit from statins (Figure S1)19, and that disclosure of such a GRS to individuals influences statin initiation 20. One study assessed the impact of direct-to-consumer CHD susceptibility genotypes on information seeking and sharing 21. In that study, participants appreciated the non-deterministic contributions of their genetics and behavior to CHD risk. Most individuals searched online for information about the effect of health habits and family history on CHD risk, and also discussed their CHD genotype risk information with their spouses, family members, and health care providers. While this study provided a report on direct-to-consumer disclosure of genotypic information, the influence of disclosing a multi-locus GRS for CHD on information seeking and sharing behavior has yet to be studied.

In a post-hoc analysis of the MI-GENES study, we hypothesized that disclosure of a GRS for CHD would increase information seeking and sharing with others in patients’ biological and social networks. We also hypothesized that this would differ between individuals with high GRS and those with low GRS.

Materials and Methods

MI-GENES Study

The MI-GENES study has been recently reported 20 (Figure 1). Informed consent was obtained from all subjects. The study was approved by the Mayo Clinic Internal Review Board. Trial participants (n=203) were white residents of Olmsted County, Minnesota, aged 45–65 years, not on statins, at intermediate risk for CHD (5–20%, 10-year Framingham risk score) (Table 1). Computer-generated randomization assigned patients (1:1) to the ‘CRS group’ or the ‘GRS group’ by stratifying for age, sex, and family history for CHD 20. The study was not blinded.

Figure 1.

Figure 1

Open in a new tab

Study Design. A, Of ~30,000 individuals available in the Mayo Clinic BioBank, ~2,000 met the screening criteria. A random 1,000 were selected, with 966 successfully genotyped. After targeted recruitment of at least 110 individuals with high GRS and 110 with average/low GRS, 216 participants enrolled, then 9 withdrew. Thus, 207 were randomized to the CRS and GRS groups; 203 remained at follow-up. Information seeking (circle) was assessed at baseline and at three and six months post-disclosure. Information sharing (triangle) was assessed at three and six months post-disclosure. CHD = coronary heart disease; CRS = conventional risk score; GRS = genetic risk score.

Table 1.

Baseline sociodemographic characteristics (n=203)

CRS n=100 GRS n=103 P-value
Age (years) 59.4±5.3 59.4±4.9 0.97
Female sex, n (%) 51 (51.0%) 55 (53.4%) 0.84
Family history of CHD, n (%) 30 (30.0%) 25 (24.3%) 0.45
College education*, n (%) 67 (67.0%) 58 (56.3%) 0.16
GRS 1.11±0.30 1.14±0.29 0.54
CRS, 10-year probability 8.48±3.76 8.56±4.47 0.88
Open in a new tab
*

Some college education, a college degree, or higher.

CHD: coronary heart disease; CRS: conventional risk score; GRS: genetic risk score.

Baseline lipid levels were measured, and low-density lipoprotein cholesterol (LDL-C) levels were estimated. Three months later, risk was disclosed by a genetic counselor in a standardized session. Participants randomized to the CRS group received their CRS, which was discussed with the genetic counselor, along with a brief review of family history. Patient participants randomized to the GRS group received their GRS and a combined genetic and conventional risk score (GRS multiplied by CRS). The score interpretation was discussed with the genetic counselor, along with a brief review of family history and discussion of the probabilistic nature of the GRS. In addition, the genetic counselor recommended dietary and lifestyle modifications that aim to reduce the participants’ CHD risk, in both the CRS and GRS groups. This was followed by shared decision-making (SDM) with a physician to determine the need for statin therapy 22.

Study physicians utilized a web-based clinical decision support tool – a version of the Mayo Clinic Statin Choice Decision Aid (http://migenesstudy.mayoclinic.org, password: migenes) modified for research purposes to include the GRS used in MI-GENES. Patients were shown a pictograph demonstrating their percentage CRS, and then the genome-informed risk after multiplication of the CRS by the GRS, if randomized to receive their GRS. The pictograph depicted 100 people “like the participant” and indicated how many could be expected to experience an adverse CHD event over the next 10 years. We also assessed study participants’ perception of the SDM process and satisfaction with the physician encounter, and used an established OPTION questionnaire while viewing the videotaped SDM patient-physician office visits, and demonstrated that there was no difference in the perception or quality of SDM between the two groups 22.

At three and six months post-disclosure, LDL-C levels were determined, with results placed in the EHR. The primary endpoint of the trial was change in LDL-C at six months post-disclosure, and secondary endpoints included statin initiation and changes in fat intake or physical activity 20. Psychosocial parameters were also addressed, including information seeking and sharing.

Information seeking and sharing surveys

The majority of survey questions were adapted from Health Information National Trends (HINTS) Surveys (http://hints.cancer.gov), and other established surveys 21,23–27 (Table S1).

The ‘Internet Use’ survey was administered at three and six months post-disclosure (Tables 2 and S1), using a Likert scale of 1 for “No”, 2 for “Yes”, and 3 for “Don’t know” (Table S2). Given the low frequency of responses for “No” and “Don’t know”, the responses “No” and “Don’t know” were considered unfavorable, while “Yes” was considered favorable, for conversion to a binary scale with a score of 1 for unfavorable and 2 for favorable. Questions regarding internet use for CHD information were adapted from HINTS 2012. The question, “Have you used the Internet for the following?” was adapted from the following HINTS survey years: “looked for health or medical information” (2011), “visit an Internet web site to learn specifically about CHD?” (2008), “used a website to help you with your diet, weight, or physical activity?” (2005). The questions “Have you used the Internet for any of the following: used the Internet to communicate with a doctor’s office; kept track of personal health information?” were also adapted from HINTS 2008 and HINTS 2005, respectively, and were considered internet use for PHR access.

Table 2.

Information seeking behaviors post-disclosure

3 months post-disclosure 6 months post-disclosure
Information Seeking: via websites or patient portals CRS n=100 (%) GRS n=103 (%) P-value CRS n=100 (%) GRS n=103 (%) P-value
Have you looked for information about how personal health habits, such as your diet and exercise, affect CHD risk? 56 (56) 64 (62) 0.61 37 (37) 47 (46) 0.17
Have you looked for information about how genetic factors affect CHD risk? 39 (39) 32 (31) 0.15 15 (15) 27 (26) 0.04
Do you have access to your Patient Portal? 78 (78) 84 (82) 0.35 71 (71) 77 (75) 0.46
If yes, did you sign-up for the Patient Portal after study enrollment? 4 (4) 5 (5) 0.82 5 (5) 11 (11) 0.21
Did you use the Patient Portal to access information related to your CHD risk during this study? 15 (15) 17 (17) 0.83 10 (10) 25 (24) 0.01
After study enrollment, did you search for “Direct-to-Consumer” genetic testing websites? 0 (0) 0 (0) 1.00 0 (0) 3 93) 0.25
Internet Use: Online activity and electronic communication CRS n=100 (%) GRS n=103 (%) P-value CRS n=100 (%) GRS n=103 (%) P-value
In the past 3 months, have you used the Internet to look for CHD information for yourself? 17 (17) 17 (17) 0.90 8 (8) 21 (20) 0.01
Is there a specific Internet site you like to go to for CHD information? 9 (9) 16 (16) 0.17 5 (5) 10 (10) 0.21
In the last 3 months, have you used the Internet for any of the following?
 A. Used e-mail/Internet to communicate with a doctor’s office? 48 (48) 45 (44) 0.74 41 (41) 47 (46) 0.33
 B. Looked for health/medical information? 64 (64) 68 (66) 0.66 61 (61) 71 (69) 0.21
 C. Used a website to help with diet, weight, or physical activity? 36 936) 44 (43) 0.31 28 (28) 36 (35) 0.24
 D. Kept track of personal health information? 63 (63) 62 (62) 0.71 56 (56) 63 (61) 0.25
 E. Do anything else health-related on the Internet? 26 (26) 39 (38) 0.06 28 (28) 41 (40) 0.05
 F. Visit a web site to learn specifically about CHD? 10 (10) 20 (19) 0.07 4 (4) 14 (14) 0.01
Open in a new tab

Numbers/percentages for the response ‘Yes’ are shown. CHD: coronary heart disease; CRS: conventional risk score; GRS: genetic risk score.

The ‘Information Seeking’ survey was completed at three and six months post-disclosure, and used a Likert scale of 1 for “No”, 2 for “Yes”, and 3 for “Not applicable” (Tables S1 and S3). Given the low frequency of responses for “No” and “Not applicable”, these were considered unfavorable and given a score of 1, while “Yes” was considered favorable and given a score of 2. Questions about the effects of personal habits and genetics on CHD risk were adapted from Kaphingst et al 21 and HINTS 2005, respectively. A question about direct-to-consumer genetic testing websites was adapted from HINTS 2008. The following questions were included to gather data about participants’ PHR (Figure S2): “Do you have access to your Patient Portal?”, “Did you sign-up for the Patient Portal after enrollment in this study?”, and “Did you use the Patient Portal to access information related to your CHD risk?”.

The ‘Information Sharing’ survey was administered at three and six months post-disclosure (Tables S1 and S4). The questions “Have you discussed your CHD risk with others?” and “Who did you talk to about your results: friends, family members, co-workers, other?” were adapted from Kaphingst et al 21 (Table 3). The first question used a Likert scale of 1 for “Not at all”, 2 for “Very few”, 3 for “Some”, 4 for “A fair number”, and 5 for “Frequently”. For the second survey question, 1 point was given for each of 4 spheres of influence in the participant’s information sharing radius (Figure S3). For the remaining questions in the survey, a Likert scale of 1 for “No”, 2 for “Yes”, and 3 for “Not applicable” was used. Given the low frequency of responses for “No”, “Not Applicable”, Not at all” and “Very few”, these were considered unfavorable, while “Some”, “A fair number”, “Frequently”, and “Yes” were considered favorable, with a score of 1 for unfavorable and 2 for favorable. The following question was added to expand on participants’ responses regarding those with whom they discussed their CHD risk: “Did you share your CHD risk with your parents, siblings, spouse, children, extended family, or PCP)?”. The question “Have you encouraged others to be screened for risk of having a heart attack?” was also added.

Table 3.

Information sharing post-disclosure

3 months post-disclosure 6 months post-disclosure
Information Sharing CRS n=100 (%) GRS n=103 (%) P-value CRS n=100 (%) GRS n=103 (%) P-value
Have you discussed your CHD risk with others? 77 (77) 93 (90) 0.02 76 (76) 93 (90) 0.01
Who did you talk to about your results?
 A. Friends 28 (28) 36 (34) 0.30 28 (28) 40 (38) 0.10
 B. Family members 83 (83) 97 (94) 0.02 84 (84) 91 (88) 0.26
 C. Co-workers 13 (13) 22 (21) 0.08 10 (10) 21 (20) 0.03
 D. Other 2 (2) 3 (3) 0.67 2 (2) 2 (2) 0.31
Did you share your CHD risk with your:
  Parents? 18 (18) 18 (18) 0.95 19 (19) 23 (22) 0.52
  Siblings? 35 (35) 48 (46) 0.10 37 (37) 54 (52) 0.03
  Spouse? 76 (76) 79 (77) 0.88 73 (73) 80 (78) 0.36
  Children? 47 (47) 54 (52) 0.46 46 (46) 52 (51) 0.54
  Extended family? 7 (7) 13 (13) 0.17 5 (5) 16 (16) 0.01
Did you share or intend to discuss your results with your primary care provider? 63 (64) 77 (75) 0.07 59 (59) 76 (74) 0.03
Did you share your results on Facebook? 0 (0) 0 (0) 1.00 1 (1) 1 (1) 0.69
Did you share your results on Twitter? 0 (0) 0 (0) 1.00 0 (0) 0 (0) 1.00
Did you share your results on other social networking services? 0 (0) 0 (0) 1.00 1 (1) 1 (1) 0.82
Have you encouraged others to be screened for their CHD risk? 47 (47) 61 (59) 0.13 47 (47) 61 (59) 0.10
 Social Network CRS n=100 (%) GRS n=103 (%) P-value CRS n=100 (%) GRS n=103 (%) P-value
Do you have friends or family members that you talk to about your health? 83 (83) 90 (87) 0.25 76 (76) 88 (85) 0.08
Do any community organization(s) provide you with health information? 22 (22) 23 (23) 0.43 19 (19) 25 (24) 0.15
Open in a new tab

Numbers/percentages for the response ‘Yes’ are shown. CHD = coronary heart disease, CRS = conventional risk score, GRS = genetic risk score

The ‘Social Network’ survey was completed at baseline and three and six months post-disclosure (Tables S1 and S5). The survey used a Likert scale of 1 for “no”, 2 for “yes”, and 3 for “Don’t know”. Given the low frequency of responses for “No” and “Don’t know”, these were considered unfavorable and given a score of 1, while “Yes” was considered favorable and given a score of 2. The questions “Do you have friends or family members that you talk to about your health?” and “Do any community organization(s) provide you with information on health?” were adapted from HINTS 2005 (Table 3) (also see ‘Assessment of Social Media Responses’ in the Supplemental Material).

Statistical methods

Survey data were extracted from the Research Electronic Data Capture 28 software. Data analyses were conducted using JMP V9.0.2 (SAS Institute Inc., Cary, NC). Missing data were imputed using the mode for that survey question in the CRS or GRS group in which the missing data appeared. Survey results were compared in single visit survey responses and also between visits to investigate changes in survey responses over time. Due to low frequencies of unfavorable responses among participants, each Likert scale was converted to a binary scale with a score of 1 for unfavorable and 2 for favorable responses, as described in the preceding section. Scores were calculated for individual survey questions, with a higher number indicating a more favorable response. GRS was stratified as high (≥1.1) or low/average (<1.1) risk. We assessed whether information exchange differed by GRS versus CRS or by high versus low GRS disclosure. Logistic regression was used to estimate the effect of the randomized group on the binary score for each survey question. Ordered logistic regression was used to estimate the effect of the randomized group on the ordinal sharing radius (Figure S3). Data were adjusted for the following baseline socioeconomic demographics: age, sex, family history of CHD, and level of education (Table 1). These demographic characteristics were also investigated as potential predictors using multivariate analyses, 29,30 for survey questions with responses significantly different between the GRS and CRS groups (see ‘Multivariate Logistic Regression Assessing Potential Predictors’ in Supplemental Material) (Table S6). Data were also adjusted for baseline CRS and GRS, and were expressed as odds ratio with confidence interval or mean with standard error. Statistical significance was accepted as P-value of <0.05.

Post-hoc power analysis

A post-hoc power analysis was performed for each survey question using an online application (located at http://clincalc.com/stats/Power.aspx). The obtained sample sizes and responses for each group was used to determine the power for an alpha of 0.05, assuming that the observed effect size based on the responses in the CRS group and the GRS group in the study sample was equivalent to the unmeasured effect size of the corresponding greater population.

Results

Participants in the two study groups had similar sociodemographic characteristics (Table 1); all data were adjusted for these sociodemographic characteristics, using multivariate logistic regression. Notably, most (60%–70%) trial participants had some college education, a college degree, or higher. Baseline survey responses were also similar between the two groups (see ‘Baseline Survey Parameters’ in Supplemental Material) (Table S7). Overall, following risk disclosure GRS group participants were more likely than CRS group participants to seek health information online and in their PHR, and were more likely to share their CHD risk information with others, as described in the following sections.

Information Seeking: Internet Use Outside Of The PHR

GRS participants trended towards being more likely than CRS participants to visit a website specifically to learn about CHD (OR 2.17 (0.95–5.3), P=.07) and to use the internet for additional health-related reasons (OR 1.79 (0.97–3.35), P=.06), at three months post-disclosure (Table 2). This was significant at six months post-disclosure: GRS participants were more likely than CRS participants to visit a website specifically to learn about CHD (OR 4.88 (1.55–19.13), P=.01) and to report using the internet for information about CHD (OR 3.22 (1.37–8.33), P=.01) (Table 2). CRS participants trended towards being more likely than GRS participants to use the internet for information about how genetic factors affect CHD risk (OR 2.02 (0.9–4.65), P=.09) at three months post-disclosure, but at six months post-disclosure GRS participants were more likely than CRS participants to use the internet for information about how genetic factors affect CHD risk (OR 2.11 (1.03–4.47), P=.04) (Table 2). Accordingly, GRS participants were more likely than CRS participants to maintain internet use for information about how genetic factors affect CHD risk (OR 2.78 (1.46–5.45), P=.002) and to show an increase in internet use for information about CHD (OR 2.69 (1.23–6.28), P=.02), from three to six months post-disclosure (Figure 2a) (Table S8).

Figure 2.

Figure 2

Open in a new tab

Information seeking and sharing. a, Internet use and EHR/PHR access at baseline and/or 3 and 6 months after risk disclosure, with significant p-values (*P<0.05) for the mean difference between values at 3 and 6 months post-disclosure; shared decision-making for statin initiation and documentation of subsequently lowered LDL-c levels in the chart are completed by or at 3 months post-disclosure, prior to observation of significant changes in information seeking and sharing survey responses between the CRS and GRS groups from 3 to 6 months post-disclosure. b, Information sharing at 3 months post-disclosure, with significant p-values (*P<0.05) for the mean difference between values at 3 and 6 months post-disclosure. c, Sharing radius (see Figure S3) at 3 months post-disclosure, with distribution skewed towards a maximum score Σ=4 for GRS participants. d, The time course of statin use and lowering of LDL-C levels in CRS and GRS participants potentially mirrors significant changes in information seeking and sharing; LDL-C decreased from baseline significantly more in GRS participants, due to higher statin use relative to CRS participants. CHD = coronary heart disease; CRS = conventional risk score; EHR = electronic health record; GRS = genetic risk score; LDL-C = serum calculated low density lipoprotein; PHR = personal health record.

Information Seeking: PHR Access

At three months post-disclosure, there was no difference between CRS and GRS participants regarding using the Patient Portal to access CHD risk, having access to the Mayo Clinic Patient Portal, signing up for the Patient Portal, or using the internet to communicate with a doctor’s office or keep track of personal health information (Table 2). However, this increased over time in the GRS group, such that at six months post-disclosure GRS participants were more likely than CRS participants to use the Patient Portal to access their CHD risk (OR 2.99 (1.35–7.04), P=.01) (Figure 2a) (Table 2).

Interestingly, H-GRS individuals were more likely than L-GRS individuals to keep track of personal health information at three months post-disclosure (OR 4.25 (1.54–12.1), P=.005) and this persisted at six months post-disclosure (OR 3.31 (CI 1.19–9.48), P=.02).

Information Sharing

At three months post-disclosure, GRS participants were more likely than CRS participants to discuss their CHD risk with others (OR 2.7 (1.2–6.41), P=.02), particularly family members (OR 3.31 (1.26), P=.02) (Figure 2b) (Table 3). GRS participants trended towards being more likely than CRS participants to discuss their CHD risk with co-workers (OR 1.97 (0.92–4.36), P=.08) and to share or intend to share CHD risk results with their PCP (OR 1.75 (0.95–3.3), P=.07) at three months post-disclosure (Table 3). Consequently, GRS participants were more likely than CRS participants to have a wider sharing radius (OR 1.77 (1.02–3.06), P=.04) (Figure 2c) (see Figure S3). The wider sharing radius in GRS participants persisted at six months post-disclosure (OR 1.77 (1.03–3.03), P=.007), due to significantly more likely information sharing with siblings (OR 1.92 (1.06–3.51), P=.03), co-workers (OR 2.42 (1.09–5.76), P=.03), and others (OR 3.13 (1.41–7.47), P=.01), including their PCP (OR 2 (1.08–3.75), P=.03) than CRS participants (Figure 2b) (Table 3).

Of note, H-GRS individuals were more likely than CRS individuals to share or intend to discuss their CHD risk with their PCP at three months post-disclosure (OR 2.75 (1.18–6.77), P=.02) and six months post-disclosure (OR 2.39 (CI 1–5.99), P=.048). Conversely, L-GRS individuals were more likely than H-GRS individuals (OR 20.13 (2.42–458.4), P=.004) and CRS individuals (OR 4.69 (1.48–17.05), P=.01) at three months post-disclosure and remained more likely than CRS individuals (OR 6.1 (1.88–23.52), P=.002) at six months post-disclosure to share their CHD risk with extended family members. L-GRS participants were also more likely than H-GRS participants (OR 10.16 (CI 1.26–118.43), P=.03) and CRS participants (OR 4.08 (0.79–25.85), P=.09) to encourage others to be screened for their CHD risk at three months post-disclosure.

Social Network

There was no significant difference in social network between the GRS and CRS participants at three months post-disclosure (Table 3). GRS participants trended towards having more friends or family members with whom they discussed their health than CRS participants at six months post-disclosure (OR 1.92 (0.92–4.11), P=.08). H-GRS individuals were more likely than CRS individuals to report having friends or family members with whom they discussed health at six months post-disclosure (OR 3.24 (1.13–10.85), P=.03).

Post-hoc power analysis

The MI-GENES Study was not specifically powered to assess information seeking and sharing. Consistent with this, a post-hoc power analysis suggested that the study was overall underpowered (β<0.80 for each survey response) for this analysis. Regardless, several statistically significant differences were noted between the CRS and GRS group, revealing salient patterns of information seeking and sharing between the two groups. The effect size for each of several survey responses was described in the previous sections.

Discussion

Disclosure of CHD risk estimates that included a GRS led to increased information seeking and information sharing behaviors in the MI-GENES trial. Early on in the study, CRS participants trended towards being more likely to search for information about the effect of genetic factors on CHD risk, likely responding to personal curiosity and being randomized to the group not receiving genetic risk information. Genetic risk information was not discussed with CRS participants. However, the impact of family history on CHD risk was briefly reviewed. All study participants were made aware at their first study visit that they would receive either a CRS or a GRS combined with their CRS at the second study visit. CRS participants were also informed that they would receive their genome-informed risk at the end of the trial. This may have enticed interest in seeking information online regarding the impact of genetics on CHD risk in a somewhat comparable fashion to the participants who received their genome-informed risk early on in the study. Over time, there was an increase among GRS participants in seeking information related to the effect of genetics on CHD risk, while this behavior decreased in CRS participants.

Patients who received their GRS early on used the internet to seek health information and accessed their PHR through the patient portal more often than patients who received their CRS alone. This trend did not reach significance at three months post-disclosure, but became significant at six months post-disclosure. This suggested that as the trial advanced, GRS participants were more interested in returning to retrieve their score. It is possible that GRS participants sought more information regarding genetics and CHD risk from websites online after observing a significantly lower LDL-C level (than baseline, after significantly higher initiation of statin therapy than CRS participants at three months post-disclosure 20) (Figure 2d) placed in the EHR immediately following the visit at three months post-disclosure. This could contribute to increased information seeking from three to six months post-disclosure, supporting the important role of information seeking in the process of coping with health concerns 31.

Overall, individuals with high GRS exhibited even higher information seeking and sharing behaviors than L-GRS and CRS participants for various parameters, such as accessing health records to keep track of personal health information, sharing CHD risk with their PCP, and having friends and family with whom they discussed health, suggesting that a high GRS can assist with health promotion. On the other hand, L-GRS individuals were more likely to share their CHD risk with extended family members and encourage others to be screened for their CHD risk, likely reflecting their reassurance from the low risk result. These results suggest that in different ways, both high and low genetic risk can influence information seeking and sharing.

To facilitate informed decision-making in precision medicine, we will need to ensure accurate sources of information online for patients, biological and social networks, physicians, and other healthcare workers 1. Patients often present to their doctors with health information found online that is incorrect or of poor quality, and with reports that useful health information is difficult to find electronically 1,23. Providing reliable information of good quality that is easy to find online will aid engagement of patients in precision medicine, as they seek counsel from family and social networks to make decisions for their personalized care 3. We provide a carefully selected list of internet websites identified by study participants or recommended by our group for patient and provider education, along with further discussion of enhancing the GRS report for both patient and provider education (see ‘Internet Websites’ and ‘Enhancing GRS Reports’ in Supplemental Material).

GRS disclosure early on led to increased information sharing with others (Figure S3), particularly family members, friends, coworkers, and PCPs. Patients likely shared their CHD genetic risk information (above and beyond sharing of non-genetic risk information) with family members for social support and specifically due to anticipated shared risk, as found in other studies. 32 Participants also shared their CHD risk information with others in their non-biological social networks, most likely to obtain emotional support from friends and coworkers, and to aid continuity of care with their PCP. This would be consistent with studies suggesting that individuals rate their friends and family as their top source of emotional support, and health professionals as their top source for technical health issues such as diagnosis and treatment. 33

By patients sharing CHD risk information, there is potential for cooperative strategies to adopt healthy lifestyles, or to make decisions for medication use to lower CHD risk, corresponding to a communal coping in social networks hypothesis (see Figure S4) 10. Our data therefore suggest that participants themselves can potentially have an impact on their community by sharing this health information. Indeed, network-based approaches have been shown to optimize population-level behavior changes 34, and should be explored in response to GRS disclosure. In addition, ongoing studies in our group are assessing participants’ perceptions of experience, salience, beliefs, risk, knowledge, subjective norms, self-efficacy, locus of control, outcome expectations, attitudes, and social networks that fit into the Comprehensive Model of Information Seeking (CMIS), expanded Planned Risk Information Seeking Model (PRISM), Health Information Model, and other models of information behavior 35–38 that can potentially be applied to social network medicine.

Information seeking and sharing behaviors in the MI-GENES study are more similar to behaviors noted in other genetic studies related to CHD and its risk factors than to cancer-related or non-genetic studies (see ‘Information seeking and sharing compared to other genetic and non-genetic studies’ in Supplemental Material). This is most likely due to perception of risk for various cancers as non-modifiable, while CHD risk is considered modifiable. Such behaviors include individuals who have received genetic risk information for CHD exhibiting high levels of seeking information online related to their test results or the effect of heritability or health habits on disease risk, and sharing risk information with others including their PCPs, family, and friends 21. While the effect on information seeking and sharing reported here may not be exclusive to disclosure of a GRS for CHD, examining this relationship can be informative for CHD prevention and health promotion.

Interestingly, participants in the current study shared their risk information most with spouses, then with siblings or children, and least with parents. In contrast, in a study of heritable cancers, individuals shared their risk information most with their parents and least with their children 39. This difference could be due to individuals who have received cancer genetic test results protecting their children from traumatic information 40, while individuals with CHD genetic risk information may share this readily with their spouses, siblings, and children, not only due to shared risk but also due to the potential for communal coping and risk modification. Sharing risk information with non-biological contacts has been reported in studies disclosing genetic risk for CHD and its risk factors (including this study), and also to disclosure of non-genetic medical information 41, but this is not a frequent occurrence in cancer genetic risk studies 39,40.

While a meta-analysis inferred that disclosure of genetic risk estimates may not prompt behavioral changes (e.g., diet, physical activity, smoking), the majority of studies appraised disclosed only a single genotype primarily for cancers and risk factors for CHD (e.g., obesity, hypertension, familial hypercholesterolemia, and type 2 diabetes) 42. No study disclosed a multi-locus GRS for CHD, and information seeking and sharing were not assessed. One study revealed changes in medication use when comparing participants who received positive versus negative risk genotype results for Alzheimer’s disease. In another study, individuals who received genetic results displayed increased perception of medication efficacy, implying potential willingness to initiate medication had it not already been prescribed 43. Several studies have suggested modest clinical utility of a GRS for CHD 12–19, 44,45. In our report of the primary outcomes of the MI-GENES randomized clinical trial, we presented evidence for significantly higher rates of medication use (i.e., statin initiation) in those who received their probabilistic GRS for CHD relative to conventional risk estimates alone, with a similar trend in those with high versus low GRS 20. Thus, disclosure of a GRS for CHD may modify preventive medication use, in addition to information behavior promoting health (see ‘Potential Impact of Information Behavior on Decision-Making’ in Supplemental Material).

This analysis of the MI-GENES trial is the first study of the impact of EHR-based CHD GRS disclosure on information seeking and sharing. Strengths of the study include standardized genetic counseling and SDM with a physician in a randomized controlled trial. The post-hoc power analyses indicated that the study was underpowered for information seeking and sharing survey questions compared between the CRS and GRS groups, but the study revealed various statistically significant differences between the two groups, supporting the main theses of our results. Further studies focusing on information seeking and sharing should be adequately powered a priori to uncover additional trends and confirm data from the current study. Individuals may have been disappointed after randomization to the CRS group, which could influence their self-reported survey responses. The MI-GENES trial design was completed prior to publication of the ACC/AHA 2013 recommendations for blood cholesterol management 46. As a result, the Framingham risk score was used in the MI-GENES study, predating the ASCVD pooled cohort equations. Beyond office visits, the PHR, and educational handouts, our study did not assess the quality of information sought or shared online or in participants’ social networks.

Study participants were overall well-educated individuals of European ancestry living in Olmsted County in Minnesota; 60%–70% of study participants reported some college education, a college degree, or higher. While limiting generalizability, the education level of participants in this study is similar to previous epidemiological reports for Olmsted County 47, as well as the national average for the general population 48 and populations reported in a myriad of other studies on health information seeking (see ‘Potential impact of sociodemographic characteristics on Information Behavior’ in Supplemental Material). Education and other sociodemographic characteristics are just one part of numerous parameters that shape information seeking. After adjustment for these sociodemographic characteristics, GRS disclosure led to further increase in information seeking and sharing.

Supplementary Material

001613 - Supplemental Material

Clinical Perspective.

Whether disclosure of a genetic risk score for the prevention of a common disease influences patients’ information seeking and sharing is unknown. In this manuscript, we report on information seeking and sharing after disclosure of genetic risk for coronary heart disease (CHD). Overall, trial participants who received their genetic risk score were more likely than those who received conventional risk information alone to use the internet for learn about CHD and seek information about how genetic factors affect CHD risk, as well as use the patient portal to access their CHD risk and discuss their CHD risk with others. This elucidates the need for continued discussions on optimizing information available to patients online and targeting information sharing by patients for health promotion. To the best of our knowledge, this is the only genomic medicine clinical trial that includes both disclosure of a multi-locus genetic risk score for coronary heart disease and assessment of individuals’ information management behaviors. Elucidation of such behaviors is relevant to patient engagement in precision medicine. The manuscript highlights information seeking and sharing, which are psychosocial responses, by individuals after receiving results of multi-locus genetic risk testing in cardiovascular disease. Clinicians may apply these results to their patient base, if sociodemographic characteristics are similar.

Acknowledgments

We are grateful to Dr. Erin Austin at Mayo Clinic in Rochester, MN for assistance with statistical analysis.

Sources of Funding: This study was part of the NHGRI-funded supported eMERGE (Electronic Records and Genomics) Network (U01HG04599 and U01HG006379). The Mayo Clinic Biobank was funded by the Mayo Clinic Center for Individualized Medicine. Use of REDCap was funded by the Center for Clinical and Translational Science grant support (UL1TR000135). Approved by: Mayo Clinic Institutional Review Board on 6-28-2013.

Footnotes

Disclosures: None.

References

  • 1.Hood L, Auffray C. Participatory medicine: a driving force for revolutionizing healthcare. Genome Med. 2013;5:110. doi: 10.1186/gm514. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.deBronkart D. From patient centred to people powered: autonomy on the rise. BMJ. 2015;350:h148. doi: 10.1136/bmj.h148. [DOI] [PubMed] [Google Scholar]
  • 3.Collins H, Calvo S, Greenberg K, Forman Neall L, Morrison S. Information Needs in the Precision Medicine Era: How Genetics Home Reference Can Help. Interact J Med Res. 2016;5:e13. doi: 10.2196/ijmr.5199. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Daack-Hirsch S, Campbell C. The role of patient engagement in personalized healthcare. Personalized Medicine. 2014;11:1–4. doi: 10.2217/pme.13.102. [DOI] [PubMed] [Google Scholar]
  • 5.Greene J, Hibbard JH, Sacks R, Overton V, Parrotta CD. When patient activation levels change, health outcomes and costs change, too. Health Aff (Millwood) 2015;34:431–7. doi: 10.1377/hlthaff.2014.0452. [DOI] [PubMed] [Google Scholar]
  • 6.Nelson DE, Kreps GL, Hesse BW, Croyle RT, Willis G, Arora NK, et al. The Health Information National Trends Survey (HINTS): development, design, and dissemination. J Health Commun. 2004;9:443–60. doi: 10.1080/10810730490504233. discussion 81–4. [DOI] [PubMed] [Google Scholar]
  • 7.Truog RD. Patients and doctors--evolution of a relationship. N Engl J Med. 2012;366:581–5. doi: 10.1056/NEJMp1110848. [DOI] [PubMed] [Google Scholar]
  • 8.Steehler KR, Steehler MK, Pierce ML, Harley EH. Social media’s role in otolaryngology-head and neck surgery: informing clinicians, empowering patients. Otolaryngol Head Neck Surg. 2013;149:521–4. doi: 10.1177/0194599813501463. [DOI] [PubMed] [Google Scholar]
  • 9.McBride CM, Koehly LM, Sanderson SC, Kaphingst KA. The behavioral response to personalized genetic information: will genetic risk profiles motivate individuals and families to choose more healthful behaviors? Annu Rev Public Health. 2010;31:89–103. doi: 10.1146/annurev.publhealth.012809.103532. [DOI] [PubMed] [Google Scholar]
  • 10.Koehly L, Loscalzo A. Adolescent Obesity and Social Networks. Prev Chronic Dis. 2009;6:A99. [PMC free article] [PubMed] [Google Scholar]
  • 11.The top 10 causes of death. 2014 http://www.who.int/mediacentre/factsheets/fs310/en/
  • 12.Ganna A, Magnusson PK, Pedersen NL, de Faire U, Reilly M, Arnlöv J, et al. Multilocus genetic risk scores for coronary heart disease prediction. Arterioscler Thromb Vasc Biol. 2013;33:2267–72. doi: 10.1161/ATVBAHA.113.301218. [DOI] [PubMed] [Google Scholar]
  • 13.Thanassoulis G, Peloso GM, Pencina MJ, Hoffmann U, Fox CS, Cupples LA, et al. A genetic risk score is associated with incident cardiovascular disease and coronary artery calcium: the Framingham Heart Study. Circ Cardiovasc Genet. 2012;5:113–21. doi: 10.1161/CIRCGENETICS.111.961342. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Tikkanen E, Havulinna AS, Palotie A, Salomaa V, Ripatti S. Genetic risk prediction and a 2-stage risk screening strategy for coronary heart disease. Arterioscler Thromb Vasc Biol. 2013;33:2261–6. doi: 10.1161/ATVBAHA.112.301120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Ripatti S, Tikkanen E, Orho-Melander M, Havulinna AS, Silander K, Sharma A, et al. A multilocus genetic risk score for coronary heart disease: case-control and prospective cohort analyses. Lancet. 2010;376:1393–400. doi: 10.1016/S0140-6736(10)61267-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Brautbar A, Pompeii LA, Dehghan A, Ngwa JS, Nambi V, Virani SS, et al. A genetic risk score based on direct associations with coronary heart disease improves coronary heart disease risk prediction in the Atherosclerosis Risk in Communities (ARIC), but not in the Rotterdam and Framingham Offspring, Studies. Atherosclerosis. 2012;223:421–6. doi: 10.1016/j.atherosclerosis.2012.05.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Hughes MF, Saarela O, Stritzke J, Kee F, Silander K, Klopp N, et al. Genetic markers enhance coronary risk prediction in men: the MORGAM prospective cohorts. PLoS One. 2012;7:e40922. doi: 10.1371/journal.pone.0040922. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Samaan Z, Schulze KM, Middleton C, Irvine J, Joseph P, Mente A, et al. South Asian Heart Risk Assessment (SAHARA): Randomized Controlled Trial Design and Pilot Study. JMIR Res Protoc. 2013;2:e33. doi: 10.2196/resprot.2621. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Mega JL, Stitziel NO, Smith JG, Chasman DI, Caulfield MJ, Devlin JJ, et al. Genetic risk, coronary heart disease events, and the clinical benefit of statin therapy: an analysis of primary and secondary prevention trials. Lancet. 2015;385:2264–71. doi: 10.1016/S0140-6736(14)61730-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Kullo IJ, Jouni H, Austin EE, Brown SA, Kruisselbrink TM, Isseh IN, et al. Incorporating a Genetic Risk Score Into Coronary Heart Disease Risk Estimates: Effect on Low-Density Lipoprotein Cholesterol Levels (the MI-GENES Clinical Trial) Circulation. 2016;133:1181–8. doi: 10.1161/CIRCULATIONAHA.115.020109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Kaphingst KA, McBride CM, Wade C, Alford SH, Reid R, Larson E, et al. Patients’ understanding of and responses to multiplex genetic susceptibility test results. Genet Med. 2012;14:681–7. doi: 10.1038/gim.2012.22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Jouni H, Haddad RA, Marroush TS, Brown SA, Kruisselbrink TM, Austin EE, et al. Shared decision-making following disclosure of coronary heart disease genetic risk: results from a randomized clinical trial. J Investig Med. 2017;65:681–688. doi: 10.1136/jim-2016-000318. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Miller LM, Bell RA. Online health information seeking: the influence of age, information trustworthiness, and search challenges. J Aging Health. 2012;24:525–41. doi: 10.1177/0898264311428167. [DOI] [PubMed] [Google Scholar]
  • 24.Wilson V, Balkan S, Lankton N. Trends in U.S. Consumers’ Use of E-Health Services: Fine-Grained Results from a Longitudinal, Demographic Survey. Communications of the Association for Information Systems. 2014:34. [Google Scholar]
  • 25.Volkman JE, Luger TM, Harvey KL, Hogan TP, Shimada SL, Amante D, et al. The National Cancer Institute’s Health Information National Trends Survey [HINTS]: a national cross-sectional analysis of talking to your doctor and other healthcare providers for health information. BMC Fam Pract. 2014;15:111. doi: 10.1186/1471-2296-15-111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Beckjord EB, Finney Rutten LJ, Squiers L, Arora NK, Volckmann L, Moser RP, et al. Use of the internet to communicate with health care providers in the United States: estimates from the 2003 and 2005 Health Information National Trends Surveys (HINTS) J Med Internet Res. 2007;9:e20. doi: 10.2196/jmir.9.3.e20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Squiers L, Bright MA, Rutten LJ, Atienza AA, Treiman K, Moser RP, et al. Awareness of the National Cancer Institute’s Cancer Information Service: results from the Health Information National Trends Survey (HINTS) J Health Commun. 2006;11(Suppl 1):117–33. doi: 10.1080/10810730600637517. [DOI] [PubMed] [Google Scholar]
  • 28.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381. doi: 10.1016/j.jbi.2008.08.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Finney Rutten LJ, Agunwamba AA, Wilson P, Chawla N, Vieux S, Blanch-Hartigan D, et al. Cancer-Related Information Seeking Among Cancer Survivors: Trends Over a Decade (2003–2013) J Cancer Educ. 2016;31:348–57. doi: 10.1007/s13187-015-0802-7. [DOI] [PubMed] [Google Scholar]
  • 30.Fox CS, Hwang SJ, Nieto K, Valentino M, Mutalik K, Massaro JM, et al. Digital Connectedness in the Framingham Heart Study. J Am Heart Assoc. 2016;5:e003193. doi: 10.1161/JAHA.116.003193. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Xiao N, Sharman R, Rao HR, Upadhyaya S. Factors influencing online health information search: An empirical analysis of a national cancer-related survey. Decision Support Systems. 2014;57:417–27. [Google Scholar]
  • 32.Stoffel EM, Ford B, Mercado RC, Punglia D, Kohlmann W, Conrad P, et al. Sharing genetic test results in Lynch syndrome: communication with close and distant relatives. Clin Gastroenterol Hepatol. 2008;6:333–8. doi: 10.1016/j.cgh.2007.12.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Fox S. After Dr Google: peer-to-peer health care. Pediatrics. 2013;131(Suppl 4):S224–5. doi: 10.1542/peds.2012-3786K. [DOI] [PubMed] [Google Scholar]
  • 34.Kim DA, Hwong AR, Stafford D, Hughes DA, O’Malley AJ, Fowler JH, et al. Social network targeting to maximise population behaviour change: a cluster randomised controlled trial. Lancet. 2015;386:145–53. doi: 10.1016/S0140-6736(15)60095-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Hovick SR, Kahlor L, Liang MC. Personal cancer knowledge and information seeking through PRISM: the planned risk information seeking model. J Health Commun. 2014;19:511–27. doi: 10.1080/10810730.2013.821556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Longo DR, Schubert SL, Wright BA, LeMaster J, Williams CD, Clore JN. Health information seeking, receipt, and use in diabetes self-management. Ann Fam Med. 2010;8:334–40. doi: 10.1370/afm.1115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Ruppel EK. Scanning Health Information Sources: Applying and Extending the Comprehensive Model of Information Seeking. J Health Commun. 2016;21:208–16. doi: 10.1080/10810730.2015.1058438. [DOI] [PubMed] [Google Scholar]
  • 38.Wilson T. Models in information behaviour research. Journal of Documentation. 1999;55:249–70. [Google Scholar]
  • 39.Elrick A, Ashida S, Ivanovich J, Lyons S, Biesecker BB, Goodman MS, et al. Psychosocial and Clinical Factors Associated with Family Communication of Cancer Genetic Test Results among Women Diagnosed with Breast Cancer at a Young Age. J Genet Couns. 2017;26:173–181. doi: 10.1007/s10897-016-9995-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Gallo AM, Angst DB, Knafl KA. Disclosure of genetic information within families. Am J Nurs. 2009;109:65–9. doi: 10.1097/01.NAJ.0000348607.31983.6e. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Li N, Orrange S, Kravitz RL, Bell RA. Reasons for and predictors of patients’ online health information seeking following a medical appointment. Fam Pract. 2014;31:550–6. doi: 10.1093/fampra/cmu034. [DOI] [PubMed] [Google Scholar]
  • 42.Hollands GJ, French DP, Griffin SJ, Prevost AT, Sutton S, King S, et al. The impact of communicating genetic risks of disease on risk-reducing health behaviour: systematic review with meta-analysis. BMJ. 2016;352:i1102. doi: 10.1136/bmj.i1102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Marteau T, Senior V, Humphries SE, Bobrow M, Cranston T, Crook MA, et al. Psychological impact of genetic testing for familial hypercholesterolemia within a previously aware population: a randomized controlled trial. Am J Med Genet A. 2004;128A:285–93. doi: 10.1002/ajmg.a.30102. [DOI] [PubMed] [Google Scholar]
  • 44.Iribarren C, Lu M, Jorgenson E, Martínez M, Lluis-Ganella C, Subirana I, et al. Clinical Utility of Multi-marker Genetic Risk Scores for Prediction of Incident Coronary Heart Disease: A Cohort Study among over 51 Thousand Individuals of European Ancestry. Circ Cardiovasc Genet. 2016;9:531–540. doi: 10.1161/CIRCGENETICS.116.001522. [DOI] [PubMed] [Google Scholar]
  • 45.Khera AV, Emdin CA, Drake I, Natarajan P, Bick AG, Cook NR, et al. Genetic Risk, Adherence to a Healthy Lifestyle, and Coronary Disease. N Engl J Med. 2016;375:2349–58. doi: 10.1056/NEJMoa1605086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Stone NJ, Robinson JG, Lichtenstein AH, Bairey Merz CN, Blum CB, Eckel RH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014;129(25 Suppl 2):S1–45. doi: 10.1161/01.cir.0000437738.63853.7a. [DOI] [PubMed] [Google Scholar]
  • 47.St Sauver JL, Grossardt BR, Leibson CL, Yawn BP, Melton LJ, Rocca WA. Generalizability of epidemiological findings and public health decisions: an illustration from the Rochester Epidemiology Project. Mayo Clin Proc. 2012;87:151–60. doi: 10.1016/j.mayocp.2011.11.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Ryan CL, Bauman K. Educational attainment in the United States: 2015. Current Population Reports. 2016 Mar;:20. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

001613 - Supplemental Material
NIHMS889696-supplement-001613_-_Supplemental_Material.pdf (909.9KB, pdf)

RESOURCES