Limited-Variant Screening vs Comprehensive Genetic Testing for Familial Hypercholesterolemia Diagnosis | Cardiology | JAMA Cardiology | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  Percentage of Individuals With Clinically Significant Familial Hypercholesterolemia Variants as Detected by a Comprehensive Test and a Limited-Variant Screen
Percentage of Individuals With Clinically Significant Familial Hypercholesterolemia Variants as Detected by a Comprehensive Test and a Limited-Variant Screen

Positive detection rate (%) in the familial hypercholesterolemia indication cohort (n = 4563) and the proactive cohort (n = 6482) as detected by comprehensive gene sequencing (comprehensive test) and projected for 24 familial hypercholesterolemia variants included in a limited-variant screen (limited screen). Heterozygous familial hypercholesterolemia included findings of a single pathogenic or likely pathogenic (P/LP) variant in LDLR, APOB, or PCSK9. Homozygous familial hypercholesterolemia included 2 (homozygous or compound-heterozygous) P/LP variants in LDLR, APOB, or PCSK9. Autosomal recessive hypercholesterolemia included findings of 2 P/LP variants in LDLRAP1. Autosomal recessive hypercholesterolemia carrier included a single P/LP variant in LDLRAP1. Analyses included only gain-of-function variants in APOB and assumed that the limited-variant screen does not report zygosity.

Figure 2.  Self-reported Ancestry of Familial Hypercholesterolemia–Positive Individuals Who Would Be Missed by a Limited-Variant Screen
Self-reported Ancestry of Familial Hypercholesterolemia–Positive Individuals Who Would Be Missed by a Limited-Variant Screen

Among all individuals in the indication cohort positive for a clinically significant familial hypercholesterolemia variant, some would have been identified by both the comprehensive genetic test and a limited-variant screen (dark blue), while others would have been identified only by the comprehensive genetic test (light blue). The multiple category includes 1 or more combinations of the ancestry groups listed in the Figure or included in Other, which includes French Canadian, Mediterranean, Native American, Pacific Islander, and Sephardic Jewish.

Table 1.  Demographic Characteristics of Indication and Proactive Cohorts
Demographic Characteristics of Indication and Proactive Cohorts
Table 2.  Zygosity and Clinical Designation of Familial Hypercholesterolemia Comprehensive Test and Limited-Variant Screen Findings in the Indication Cohort
Zygosity and Clinical Designation of Familial Hypercholesterolemia Comprehensive Test and Limited-Variant Screen Findings in the Indication Cohort
Table 3.  Cascade Testing of Relatives of Individuals With a Positive Molecular Finding for Familial Hypercholesterolemia
Cascade Testing of Relatives of Individuals With a Positive Molecular Finding for Familial Hypercholesterolemia
1.
Benn  M, Watts  GF, Tybjærg-Hansen  A, Nordestgaard  BG.  Mutations causative of familial hypercholesterolaemia: screening of 98 098 individuals from the Copenhagen General Population Study estimated a prevalence of 1 in 217.   Eur Heart J. 2016;37(17):1384-1394. doi:10.1093/eurheartj/ehw028PubMedGoogle ScholarCrossref
2.
Abul-Husn  NS, Manickam  K, Jones  LK,  et al.  Genetic identification of familial hypercholesterolemia within a single U.S. health care system.   Science. 2016;354(6319):aaf7000. doi:10.1126/science.aaf7000PubMedGoogle Scholar
3.
Trinder  M, Francis  GA, Brunham  LR.  Association of monogenic vs polygenic hypercholesterolemia with risk of atherosclerotic cardiovascular disease.   JAMA Cardiol. 2020;5(4):390-399. doi:10.1001/jamacardio.2019.5954PubMedGoogle ScholarCrossref
4.
Lamiquiz-Moneo  I, Civeira  F, Mateo-Gallego  R,  et al.  Diagnostic yield of sequencing familial hypercholesterolemia genes in individuals with primary hypercholesterolemia.   Rev Esp Cardiol (Engl Ed). Published online July 10, 2020. doi:10.1016/j.rec.2020.06.003PubMedGoogle Scholar
5.
Khera  AV, Won  H-H, Peloso  GM,  et al.  Diagnostic yield and clinical utility of sequencing familial hypercholesterolemia genes in patients with severe hypercholesterolemia.   J Am Coll Cardiol. 2016;67(22):2578-2589. doi:10.1016/j.jacc.2016.03.520PubMedGoogle ScholarCrossref
6.
Tada  H, Kawashiri  M-A, Nohara  A, Inazu  A, Mabuchi  H, Yamagishi  M.  Impact of clinical signs and genetic diagnosis of familial hypercholesterolaemia on the prevalence of coronary artery disease in patients with severe hypercholesterolaemia.   Eur Heart J. 2017;38(20):1573-1579. doi:10.1093/eurheartj/ehx004PubMedGoogle ScholarCrossref
7.
Nordestgaard  BG, Chapman  MJ, Humphries  SE,  et al; European Atherosclerosis Society Consensus Panel.  Familial hypercholesterolaemia is underdiagnosed and undertreated in the general population: guidance for clinicians to prevent coronary heart disease: consensus statement of the European Atherosclerosis Society.   Eur Heart J. 2013;34(45):3478-90a. doi:10.1093/eurheartj/eht273PubMedGoogle ScholarCrossref
8.
Versmissen  J, Oosterveer  DM, Yazdanpanah  M,  et al.  Efficacy of statins in familial hypercholesterolaemia: a long term cohort study.   BMJ. 2008;337:a2423. doi:10.1136/bmj.a2423PubMedGoogle ScholarCrossref
9.
Luirink  IK, Wiegman  A, Kusters  DM,  et al.  20-Year follow-up of statins in children with familial hypercholesterolemia.   N Engl J Med. 2019;381(16):1547-1556. doi:10.1056/NEJMoa1816454PubMedGoogle ScholarCrossref
10.
Fahed  AC, Nemer  GM.  Familial hypercholesterolemia: the lipids or the genes?   Nutr Metab (Lond). 2011;8(1):23. doi:10.1186/1743-7075-8-23PubMedGoogle ScholarCrossref
11.
Sturm  AC, Knowles  JW, Gidding  SS,  et al; Convened by the Familial Hypercholesterolemia Foundation.  Clinical genetic testing for familial hypercholesterolemia: JACC Scientific Expert Panel.   J Am Coll Cardiol. 2018;72(6):662-680. doi:10.1016/j.jacc.2018.05.044PubMedGoogle ScholarCrossref
12.
Henderson  R, O’Kane  M, McGilligan  V, Watterson  S.  The genetics and screening of familial hypercholesterolaemia.   J Biomed Sci. 2016;23:39. doi:10.1186/s12929-016-0256-1PubMedGoogle ScholarCrossref
13.
23andMe. 23andMe personal genome service (PGS) package insert. Published June 2020. Accessed July 24, 2020. https://permalinks.23andme.com/pdf/package_insert_v5.pdf
14.
23andMe Blog. 23andMe offers a new report on familial hypercholesterolemia. Published February 20, 2019. Accessed July 24, 2020. https://blog.23andme.com/health-traits/23andme-offers-a-new-report-on-familial-hypercholesterolemia/
15.
Lincoln  SE, Kobayashi  Y, Anderson  MJ,  et al.  A systematic comparison of traditional and multigene panel testing for hereditary breast and ovarian cancer genes in more than 1000 patients.   J Mol Diagn. 2015;17(5):533-544. doi:10.1016/j.jmoldx.2015.04.009PubMedGoogle ScholarCrossref
16.
Nykamp  K, Anderson  M, Powers  M,  et al; Invitae Clinical Genomics Group.  Sherloc: a comprehensive refinement of the ACMG-AMP variant classification criteria.   Genet Med. 2017;19(10):1105-1117. doi:10.1038/gim.2017.37PubMedGoogle ScholarCrossref
17.
Truty  R, Paul  J, Kennemer  M,  et al.  Prevalence and properties of intragenic copy-number variation in mendelian disease genes.   Genet Med. 2019;21(1):114-123. doi:10.1038/s41436-018-0033-5PubMedGoogle ScholarCrossref
18.
Richards  S, Aziz  N, Bale  S,  et al; ACMG Laboratory Quality Assurance Committee.  Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.   Genet Med. 2015;17(5):405-424. doi:10.1038/gim.2015.30PubMedGoogle ScholarCrossref
19.
Churchwell  K, Elkind  MSV, Benjamin  RM,  et al; American Heart Association.  Call to action: structural racism as a fundamental driver of health disparities: a presidential advisory from the American Heart Association.   Circulation. 2020;142(24):e454-e468. doi:10.1161/CIR.0000000000000936PubMedGoogle ScholarCrossref
20.
eMERGE Clinical Annotation Working Group.  Frequency of genomic secondary findings among 21,915 eMERGE network participants.   Genet Med. 2020;22(9):1470-1477. doi:10.1038/s41436-020-0810-9PubMedGoogle ScholarCrossref
21.
Iacocca  MA, Wang  J, Dron  JS,  et al.  Use of next-generation sequencing to detect LDLR gene copy number variation in familial hypercholesterolemia.   J Lipid Res. 2017;58(11):2202-2209. doi:10.1194/jlr.D079301PubMedGoogle ScholarCrossref
22.
Regalado  A. More than 26 million people have taken an at-home ancestry test. MIT Technology Review. Published February 11, 2019. Accessed March 5, 2021. https://www.technologyreview.com/2019/02/11/103446/more-than-26-million-people-have-taken-an-at-home-ancestry-test
23.
Neal  WA, Knowles  J, Wilemon  K.  Underutilization of cascade screening for familial hypercholesterolemia.   Clin Lipidol. 2014;9(3):291-293. doi:10.2217/clp.14.19Google ScholarCrossref
24.
Amor-Salamanca  A, Castillo  S, Gonzalez-Vioque  E,  et al.  Genetically confirmed familial hypercholesterolemia in patients with acute coronary syndrome.   J Am Coll Cardiol. 2017;70(14):1732-1740. doi:10.1016/j.jacc.2017.08.009PubMedGoogle ScholarCrossref
25.
Ahmad  ZS, Andersen  RL, Andersen  LH,  et al.  US physician practices for diagnosing familial hypercholesterolemia: data from the CASCADE-FH registry.   J Clin Lipidol. 2016;10(5):1223-1229. doi:10.1016/j.jacl.2016.07.011PubMedGoogle ScholarCrossref
26.
Representatives of the Global Familial Hypercholesterolemia Community.  Reducing the clinical and public health burden of familial hypercholesterolemia: a global call to action.   JAMA Cardiol. 2020;5(2):217-229. doi:10.1001/jamacardio.2019.5173Google ScholarCrossref
27.
Brown  EE, Sturm  AC, Cuchel  M,  et al.  Genetic testing in dyslipidemia: A scientific statement from the National Lipid Association.   J Clin Lipidol. 2020;14(4):398-413. doi:10.1016/j.jacl.2020.04.011PubMedGoogle ScholarCrossref
28.
Musunuru  K, Hershberger  RE, Day  SM,  et al; American Heart Association Council on Genomic and Precision Medicine; Council on Arteriosclerosis, Thrombosis and Vascular Biology; Council on Cardiovascular and Stroke Nursing; and Council on Clinical Cardiology.  Genetic testing for inherited cardiovascular diseases: a scientific statement from the American Heart Association.   Circ Genom Precis Med. 2020;13(4):e000067. doi:10.1161/HCG.0000000000000067PubMedGoogle Scholar
29.
Centers for Disease Control and Prevention. More detailed information on key tier 1 applications: familial hypercholesterolemia. Published January 28, 2019. Accessed July 24, 2020. https://www.cdc.gov/genomics/implementation/toolkit/fh_1.htm
30.
Taylor  A, Wang  D, Patel  K,  et al.  Mutation detection rate and spectrum in familial hypercholesterolaemia patients in the UK pilot cascade project.   Clin Genet. 2010;77(6):572-580. doi:10.1111/j.1399-0004.2009.01356.xPubMedGoogle ScholarCrossref
31.
Haralambos  K, Whatley  SD, Edwards  R,  et al.  Clinical experience of scoring criteria for familial hypercholesterolaemia (FH) genetic testing in Wales.   Atherosclerosis. 2015;240(1):190-196. doi:10.1016/j.atherosclerosis.2015.03.003PubMedGoogle ScholarCrossref
32.
Ruhl  GL, Hazel  JW, Clayton  EW, Malin  BA.  Public attitudes toward direct to consumer genetic testing.   AMIA Annu Symp Proc. 2020;2019:774-783PubMedGoogle Scholar
33.
Savard  J, Hickerton  C, Metcalfe  SA, Gaff  C, Middleton  A, Newson  AJ.  From expectations to experiences: consumer autonomy and choice in personal genomic testing.   AJOB Empir Bioeth. 2020;11(1):63-76. doi:10.1080/23294515.2019.1701583PubMedGoogle ScholarCrossref
34.
Lee  H, Vogel  RI, LeRoy  B, Zierhut  HA.  Adult adoptees and their use of direct-to-consumer genetic testing: Searching for family, searching for health.   J Genet Couns. 2021;30(1):144-157. doi:10.1002/jgc4.1304PubMedGoogle ScholarCrossref
35.
US Food and Drug Administration. Direct-to-consumer tests. Updated December 20, 2019. Accessed July 24, 2020. https://www.fda.gov/medical-devices/vitro-diagnostics/direct-consumer-tests
36.
Tandy-Connor  S, Guiltinan  J, Krempely  K,  et al.  False-positive results released by direct-to-consumer genetic tests highlight the importance of clinical confirmation testing for appropriate patient care.   Genet Med. 2018;20(12):1515-1521. doi:10.1038/gim.2018.38PubMedGoogle ScholarCrossref
37.
Dyrbye  L, Herrin  J, West  CP,  et al.  Association of racial bias with burnout among resident physicians.   JAMA Netw Open. 2019;2(7):e197457. doi:10.1001/jamanetworkopen.2019.7457PubMedGoogle Scholar
38.
Sabin  J, Nosek  BA, Greenwald  A, Rivara  FP.  Physicians’ implicit and explicit attitudes about race by MD race, ethnicity, and gender.   J Health Care Poor Underserved. 2009;20(3):896-913. doi:10.1353/hpu.0.0185PubMedGoogle ScholarCrossref
39.
Caraballo  C, Massey  D, Mahajan  S,  et al  Racial and ethnic disparities in access to health care among adults in the United States: a 20-year national health interview survey analysis, 1999-2018.   medRxiv. Preprint published online November 4, 2020. doi:10.1101/2020.10.30.20223420Google Scholar
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    May 26, 2021

    Limited-Variant Screening vs Comprehensive Genetic Testing for Familial Hypercholesterolemia Diagnosis

    Author Affiliations
    • 1Genomic Medicine Institute, Geisinger, Danville, Pennsylvania
    • 2Invitae, San Francisco, California
    • 3Department of Genetics, Perelman School of Medicine at the University of Pennsylvania, Philadelphia
    • 4Cardiovascular Institute, Perelman School of Medicine at the University of Pennsylvania, Philadelphia
    JAMA Cardiol. Published online May 26, 2021. doi:10.1001/jamacardio.2021.1301
    Key Points

    Question  How many clinically significant variants for familial hypercholesterolemia would be missed by limited-variant screening conducted on microarrays?

    Findings  In this cross-sectional review of comprehensive genetic test results for individuals with indications for familial hypercholesterolemia, a limited-variant screen was found to have a significantly lower detection rate (8.4%) than the comprehensive diagnostic test (27%).

    Meaning  The results of this study suggest that clinically significant findings for familial hypercholesterolemia would be missed for two-thirds of affected individuals if limited-variant screening was used.

    Abstract

    Importance  Familial hypercholesterolemia (FH) is the most common inherited cardiovascular disease and carries significant morbidity and mortality risks. Genetic testing can identify affected individuals, but some array-based assays screen only a small subset of known pathogenic variants.

    Objective  To identify the number of clinically significant variants associated with FH that would be missed by an array-based, limited-variant screen when compared with next-generation sequencing (NGS)–based comprehensive testing.

    Design, Setting, and Participants  This cross-sectional study compared comprehensive genetic test results for clinically significant variants associated with FH with results for a subset of 24 variants screened by a limited-variant array. Data were deidentified next-generation sequencing results from indication-based or proactive gene panels. Individuals receiving next-generation sequencing–based genetic testing, either for an FH indication between November 2015 and June 2020 or as proactive health screening between February 2016 and June 2020 were included. Ancestry was reported by clinicians who could select from preset options or enter free text on the test requisition form.

    Main Outcomes and Measures  Number of pathogenic or likely pathogenic (P/LP) variants identified.

    Results  This study included 4563 individuals who were referred for FH diagnostic testing and 6482 individuals who received next-generation sequencing of FH-associated genes as part of a proactive genetic test. Among individuals in the indication cohort, the median (interquartile range) age at testing was 49 (32-61) years, 55.4% (2528 of 4563) were female, and 63.6% (2902 of 4563) were self-reported White/Caucasian. In the indication cohort, the positive detection rate would have been 8.4% (382 of 4563) for a limited-variant screen compared with the 27.0% (1230 of 4563) observed with the next-generation sequencing–based comprehensive test. As a result, 68.9% (848 of 1230) of individuals with a P/LP finding in an FH-associated gene would have been missed by the limited screen. The potential for missed findings in the indication cohort varied by ancestry; among individuals with a P/LP finding, 93.7% (59 of 63) of self-reported Black/African American individuals and 84.7% (122 of 144) of Hispanic individuals would have been missed by the limited-variant screen, compared with 33.3% (4 of 12) of Ashkenazi Jewish individuals. In the proactive cohort, the prevalence of clinically significant FH variants was approximately 1:191 per the comprehensive test, and 61.8% (21 of 34) of individuals with an FH-associated P/LP finding would have been missed by a limited-variant screen.

    Conclusions and Relevance  Limited-variant screens may falsely reassure the majority of individuals at risk for FH that they do not carry a disease-causing variant, especially individuals of self-reported Black/African American and Hispanic ancestry.

    Introduction

    Familial hypercholesterolemia (FH) is the most common inherited cardiovascular disease, with an estimated prevalence of 1:176 to 1:282.1-4 FH causes lifelong exposure to elevated levels of low-density lipoprotein (LDL) cholesterol that can lead to premature coronary artery disease and early death, exceeding the risks among patients with equivalent levels of LDL cholesterol due to multiple alternative factors.3,5,6 Timely pharmacological interventions can significantly reduce morbidity and mortality risks, but unfortunately most individuals with FH are undiagnosed and thus untreated.2,7-9 FH can be diagnosed clinically and via genetic testing, with the latter offering several advantages including the ability to identify affected individuals who do not meet specific cholesterol-level thresholds as well as asymptomatic and presymptomatic relatives who may be at risk for the specific disease-causing variant identified in the family.10,11 However, genetic tests vary greatly in their quality and in how many of the thousands of known and novel FH-associated genetic variants they include.

    FH is an autosomal dominant disorder most frequently associated with variants in the LDL receptor (LDLR) gene.12 Less frequently, FH is caused by variants in the apolipoprotein B (APOB) gene, primarily in the LDLR-binding region of the protein, or by gain-of-function variants in proprotein convertase subtilisin/kexin 9 (PCSK9). Additionally, a rare, recessive form of FH is caused by variants in the LDLR adaptor protein 1 gene (LDLRAP1). The particular gene affected and the type of variant can influence an individual’s level of LDL cholesterol and risk for developing coronary artery disease, as well as the intensity of pharmacologic intervention needed to control cholesterol levels. LDLR null variants are most severe, while individuals with other LDLR variants, APOB variants, and PCSK9 variants generally have a milder phenotype.2,11

    Genetic tests for FH are available in clinical settings, where a clinician will order the test on behalf of a patient, and in consumer settings, where an individual may order a test directly from a testing laboratory (often called direct-to-consumer testing). In both settings, genetic tests frequently use either array-based genotyping or next-generation sequencing (NGS) to detect variants. Critically, the technology used in a genetic test can substantially affect clinical utility. Array-based genotyping analyzes a limited set of genetic positions already known to have disease-causing variants. Sequencing-based testing is more comprehensive because it can analyze full gene sequences and detect not only known but also novel and private variants. One array-based product offered by a health and ancestry company focused on direct-to-consumer testing screens only 24 FH-associated variants as of July 2020,13 whereas tests that use NGS can detect more than 2000 known FH-associated variants as well as novel clinically actionable (pathogenic or likely pathogenic [P/LP]) variants.11

    Although most limited-variant screens are not offered in a medical or diagnostic setting, consumers may look to the results for guidance on health behaviors. Negative findings on a limited-variant screen in an individual who actually carries a disease-causing variant may lead to false reassurance that a disease-associated variant is not present, resulting in missed opportunities to initiate or intensify treatment and identify at-risk relatives. In addition, some limited-variant screens do not report whether an individual harbors 1 or 2 copies of a variant, whereas this is routine for NGS-based testing and can influence clinical management and outreach to relatives who may also need testing. Direct comparisons of these methods are needed to demonstrate how many patients with disease-associated FH variants might be identified with NGS but remain unidentified by limited-variant screens.

    Here, we performed retrospective analyses of NGS results from individuals referred for FH diagnostic testing and individuals who received NGS of FH-associated genes as part of a proactive genetic test. We determined how many individuals with positive findings of FH variants on NGS would have been missed by a limited, 24-variant screen.

    Methods

    We reviewed deidentified data under an institutional review board–approved protocol (Western Institutional Review Board #20161796) for 2 cohorts of unrelated individuals: individuals referred by an ordering clinician for FH genetic testing whose samples were received from November 2015 through June 2020 and individuals whose samples for proactive genetic testing were received from February 2016 through June 2020. Patient consent for genetic testing was obtained by the ordering clinician, who attested to this on the requisition form at the time of ordering. The proactive panel (Invitae Genetic Health Screen) was either ordered by a clinician on behalf of an individual or a request for a test was initiatied by an individual with subsequent third-party physician approval. The proactive panel included up to 147 genes addressing a broad range of inherited health conditions, including FH. To reduce potential bias toward individuals with existing cardiological health concerns, we excluded individuals who ordered the cardiology-focused proactive test (Invitae Cardio Screen). Age and ancestry (selected from preset options or entered as free text on the test requisition form) were determined by self-report (from the tested individual or their clinician); individuals with missing ancestry data were grouped as unknown in ancestry-stratified analyses. Both cohorts were tested for more than 2000 possible variants (single-nucleotide variants, insertions/deletions, and copy number variants) in 4 FH-associated genes: LDLR, APOB, PCSK9, and LDLRAP1. Because loss-of-function variants in APOB are associated with familial hypobetalipoproteinemia (a clinically distinct condition from FH), 12 loss-of-function APOB variants (none of which would be reported by the partial screen) were excluded from the analyses.

    Our projection of the limited-variant comparison assay was guided by a package insert, last updated in August 2020, which lists 24 specific single-nucleotide variants for the FH assay: 23 variants in LDLR and 1 in APOB.13 Per a February 2019 statement from the manufacturer of the comparison assay, our analysis assumed that the limited-variant assay does not report whether an individual is heterozygous or homozygous for a variant.14

    Gene Sequencing and Variant Interpretation

    Sequencing and variant classification were as previously described.15,16 Briefly, genes were targeted with oligonucleotide baits (Agilent Technologies, Roche, and Integrated DNA Technologies) to capture the exons ±10 base pairs of adjacent intronic sequence and certain noncoding regions of clinical interest. Targeted regions were sequenced to a mean depth of 350× read coverage (minimum 50×). All primary sequencing was performed on Illumina HiSeq or NovaSeq instruments (Illumina). Sequencing reads were aligned using NovoAlign (Novacraft), followed by single-nucleotide variant and small insertions/deletions variant identification using a modified GATK haplotype caller. Break points, exon-level copy number variants, and variants within homopolymers were called with additional methods.17 Resulting variants were assigned to 1 of 5 categories based on a score-based refinement of the 2015 sequence variant interpretation guidelines from the American College of Medical Genetics and Genomics and the Association for Medical Pathology: benign, likely benign, uncertain significance, likely pathogenic, or pathogenic.16,18 Of 382 unique disease-associated variants included in these analyses, 322 (84.3%) were classified as pathogenic and 60 (15.7%) were classified as likely pathogenic (eTable 1 in the Supplement).

    We grouped individuals with P/LP findings into 4 groups: (1) heterozygous FH for individuals with a single P/LP variant in LDLR, APOB, or PCSK9; (2) homozygous FH for individuals with homozygous or compound-heterozygous P/LP variants in LDLR, APOB, or PCSK9; (3) autosomal recessive hypercholesterolemia for individuals with 2 P/LP variants (homozygous or compound-heterozygous) in LDLRAP1; and (4) autosomal recessive hypercholesterolemia carrier for individuals with a single P/LP variant in LDLRAP1. Positive detection rate was defined as the proportion of individuals with heterozygous FH, homozygous FH, and autosomal recessive hypercholesterolemia among all individuals in each cohort.

    Cascade Analysis

    To compare the outcomes of cascade screening (testing of family members of affected individuals), we reviewed existing records for relatives of individuals with a P/LP result. Relatives who had testing results at the variant or gene level were grouped by proband (the individual initially included in the cohort) and assessed for the relevant FH-associated variant(s).

    Statistical Analysis

    The prop.test function in statistical software package R version 3.3.0 (R Foundation for Statistical Computing) was used to determine differences in positive detection rates between comprehensive NGS and the limited-variant screen for both the indication and proactive cohorts. Significance was set at P < .005.

    Results
    Diagnostic Testing vs Limited-Variant Screen

    The indication cohort included 4563 individuals referred by health care professionals for FH testing based on clinical suspicion. The median (interquartile range) age at testing was 49 (32-61) years, and 2528 individuals (55.4%) were female (Table 1). By self-report, 2902 (63.6%) were White/Caucasian, 315 (6.9%) were Hispanic, 280 (6.1%) were Black/African American, and 214 (4.7%) were Asian (Table 1).

    The positive detection rate by NGS was 27.0% (1230 of 4563) (eTable 2 in the Supplement). In contrast, if these individuals had been tested using the limited 24-variant array-based test, only 8.4% (382 of 4563) would have been identified and informed that they carry a disease-associated variant. Thus, 69% (848 of 1230) of individuals with P/LP variants identified in the NGS-based test would have been missed by the array-based test (Figure 1; eTable 2 in the Supplement). The performance of the limited-variant screen compared with the comprehensive test varied widely by race and ethnicity (Figure 2). The limited-variant screen would have missed 33% (4 of 12) of Ashkenazi Jewish individuals, 64% (420 of 661) of White/Caucasian individuals, 85% (122 of 144) of Hispanic individuals, and 94% (59 of 63) of Black/African American individuals positive for FH-associated variants (Figure 2).

    Additionally, NGS-based testing revealed clinically important aspects of FH genetics that may have been missed by limited-variant screening. This included 132 P/LP copy number variants that were detected by NGS in LDLR and accounted for 10.7% (132 of 1230) of the positive detection rate. Second, while the comprehensive test identified 4 individuals with a single P/LP variant in LDLRAP1 (autosomal recessive hypercholesterolemia carriers) and 9 individuals with a P/LP variant in PCSK9 (heterozygous FH), none would have been identified by the limited-variant screen because LDLRAP1 and PCSK9 are not included in the limited-variant assay (Table 2). Third, the comprehensive test identified 74 individuals with homozygous FH (Table 2). Although 37 of the individuals with homozygous FH would have received a P/LP finding from the limited-variant screen, by our analysis none would have been informed of their severe homozygous or compound-heterozygous state.

    Cascade Screening

    Given the importance of cascade (family) screening for identifying additional individuals with FH, we evaluated the uptake of cascade testing in the indication cohort and assessed the possible association of missing P/LP variants with outcomes if the limited-variant screen had been used. Among 1230 individuals with P/LP findings, 256 (20.8%) had at least 1 family member who was tested for the P/LP variant (Table 3). As a result of cascade testing, 550 family members were tested, of whom 321 individuals (58.4%) had a P/LP finding. With the limited-variant screen, only 178 additional individuals would have been tested, of whom 112 (62.9%) would have had a P/LP finding (Table 3). Thus, 65.1% of additional individuals with a P/LP FH variant would have been missed.

    Proactive Testing vs Limited-Variant Screen

    We also examined FH yields in individuals who had proactive genetic testing. Among 6482 individuals, 34 (0.5%) were found to carry at least 1 disease-associated variant in an FH-associated gene (eTable 2 in the Supplement), corresponding with a population prevalence of approximately 1:191. Only 13 of these individuals would have received findings of a P/LP variant with the limited-variant screen, indicating that 21 of the 34 individuals (61.8%) with a P/LP variant would have been missed (Figure 1). Finally, of 39 positive probands in the proactive cohort, only 2 had family members who were also tested for FH-associated genes (Table 3). Only 1 of these 2 families would have been alerted to their need for testing had the proband received results from the limited-variant screen.

    Discussion

    This study demonstrates that comprehensive NGS-based genetic testing identifies substantially more individuals affected by FH than a limited-variant screen. These results have significant implications for individuals with FH, their families, and cardiologists and other clinicians. The results suggest that two-thirds of people with a P/LP variant in a gene associated with FH would receive a negative result from a limited-variant screen that is marketed as a report of the genetic health risk for FH.14 The inferior detection rate of the limited-variant screening was most pronounced among Hispanic and Black/African American individuals. The reduced yield of limited-variant screening could result in a major health care disparity for groups already affected by social and medical disenfranchisement that beget serious health disparities including a significantly higher rate of cardiovascular death among Black/African American individuals.19

    Missed opportunities to identify individuals affected by FH could affect a relatively large portion of the population, as we found that approximately 1:191 individuals (0.52%) in the ostensibly healthy proactive cohort carried at least 1 P/LP FH variant. This prevalence estimate is nearly identical to the population prevalence of incidental findings for hereditary lipid disorders (0.5%) in a study of more than 20 000 individuals.20 In contrast with NGS-based genetic testing, the limited-variant screen evaluated in this study addresses only 24 out of thousands of potential disease-causing variants in FH-associated genes,11,13 does not report whether an individual harbors 1 or 2 copies of a variant,13,14 and does not detect copy number variants, which are reported to cause approximately 10% of FH cases.21 It should be noted that ancestry-focused direct-to-consumer tests have reportedly been used by more than 26 million individuals.22 If FH screening were part of the majority of these tests, then the absolute number of individuals receiving positive findings for FH from limited-variant screening may exceed those identified via comprehensive clinical testing. This could have substantial public health benefits. However, the challenge of negative findings remains, as the lack of genetics-informed clinician involvement with direct-to-consumer approaches may lead to misinterpretation of a negative result as a complete absence of risk for FH.

    Although identification of a P/LP variant is part of most diagnostic criteria for FH, diagnosis has traditionally been based only on clinical criteria, including elevated LDL cholesterol levels and family history of premature cardiovascular disease. However, family history may be inconclusive or unavailable and thus insensitive, and clinical observations such as cholesterol levels do not always meet diagnostic thresholds in some families with FH.2,5,23,24 Genetic testing is sufficient for diagnosis within clinical diagnostic criteria and can identify affected family members before symptom onset, enabling children to benefit from routine evaluation or treatment at a young age. A 20-year follow-up study of statin therapy in children found that none of those receiving treatment had died of cardiovascular disease before age 40 years, compared with 7% of their parents, who had initiated treatment later in life.9

    Unfortunately, genetic testing for FH is underused, with 90% of affected individuals worldwide remaining undiagnosed and only 3.9% of patients with FH in the US having a record of genetic testing.11,25,26 Recent expert statements recommend that patients suspected of having FH be offered genetic testing and that first-degree relatives of genetically positive individuals be screened for FH by lipid profile or genetic testing.11,27,28 In addition, the US Centers for Disease Control and Prevention recommends cascade screening of relatives of individuals with FH.29 While the sequencing data in this study was sourced from a single clinical laboratory, any testing facility that provides high-coverage NGS for FH genes could address the genetic testing components of these recommendations.

    Clinicians and consumers need to be aware of the limitations of any genetic test. In clinical settings, where clinicians may use limited-variant screens or comprehensive NGS tests, prior studies show that P/LP variants are identified in only 30% to 70% of individuals with a definite diagnosis of FH by LDL cholesterol levels or other clinical findings per Dutch Lipid Clinical Network scoring or Simon Broome Diagnostic Criteria for FH.24,30,31 Thus, an individual with clinical findings of FH and a negative result from any genetic test may have unidentified variants, and immediate family members remain at up to 50% risk. Further, the potential for false reassurance or incomplete information after limited-variant screening is not limited to FH. For any genetic condition with allelic heterogeneity in which multiple alleles at the same genetic locus can cause the same disease, analyzing only a limited selection of variants will likely lead to missed or incomplete diagnoses. Future work to pool comprehensive testing results from multiple testing facilities for comparison to limited-variant screening findings is needed to fully understand the disparate outcomes of various genetic technologies.

    Although most direct-to-consumer limited-variant screens are used outside of medical care, some users have reported that personal health and health-focused lifestyle changes are motivations for undergoing testing.32-34 The US Food and Drug Administration has stated that “not all direct-to-consumer genetic test companies test for the same set of variants, and therefore may provide different results for the same disease or condition.”35 Moreover, false positives have been reported for 40% of variants detected by third-party companies analyzing raw data obtained from direct-to-consumer genetic testing laboratories.36 Thus, clinical confirmation of any direct-to-consumer testing result is advised to avoid inappropriate changes in care. The US Food and Drug Administration recommends that individuals with positive direct-to-consumer testing results consult with their clinicians about next steps, which should include confirmatory clinical genetic testing.35

    Limitations

    Limitations of this study include an inability to confirm the molecular diagnoses of FH with clinical diagnoses due to the lack of medical and family histories. Although previous studies have demonstrated a 30% to 70% yield from definite FH cases,24,30,31 the yield in the indication cohort in this study was 27.0%. This should not reflect a diminished sensitivity of the NGS test because the prevalence of FH disease-associated variants in the proactive cohort was similar to frequencies found by other methods. Instead, we suspect that some of the cases in the indication cohort may have had a low a priori clinical probability of a positive test result. In addition, our assessment of limited-variant screening was a projection, as we did not have access to data from individuals who have used the comparison assay. Finally, the generalizability of our findings may be hampered by the limited representation of Hispanic and Black/African American individuals in our cohorts. This issue deserves further investigation to identify the cause or causes, which may include care, access, and coverage disparities; economic inequities (especially for the proactive cohort for which testing is largely paid out of pocket); or other reasons.37-39

    Conclusions

    Overall, our findings demonstrate a potential for significantly reduced clinical sensitivity and false reassurance among most individuals (61.8% to 68.9%) with disease-causing FH variants when limited-variant, array-based testing is used, especially among Black/African American and Hispanic individuals. Whether testing is obtained directly by a consumer or through a clinical setting, those tested should consult with a genetic counselor or other qualified health care professional to fully understand the benefits and limitations of the different types of genetic testing for FH.

    Back to top
    Article Information

    Corresponding Author: Amy C. Sturm, MS, Genomic Medicine Institute, Geisinger, Weis Center for Research, 100 N Academy Ave, Danville, PA 17822 (asturm@geisinger.edu).

    Accepted for Publication: March 25, 2021.

    Published Online: May 26, 2021. doi:10.1001/jamacardio.2021.1301

    Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License. © 2021 Sturm AC et al. JAMA Cardiology.

    Author Contributions: Dr Truty had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Ms Sturm and Dr Rader had access to all variant data that support this study.

    Concept and design: Truty, Callis, Esplin.

    Acquisition, analysis, or interpretation of data: Sturm, Truty, Callis, Aguilar, Garcia, Haverfield, Morales, Nussbaum, Rojahn, Vatta, Rader.

    Drafting of the manuscript: Morales, Rojahn.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Truty.

    Administrative, technical, or material support: Sturm, Callis, Aguilar, Haverfield, Morales, Nussbaum, Rojahn.

    Supervision: Sturm, Esplin, Garcia, Morales, Nussbaum.

    Conflict of Interest Disclosures: Ms Sturm reports that Invitae provides chatbot and genetic testing services for her National Heart, Lung, and Blood Institute–funded grant. Mss Morales and Aguilar and Drs Truty, Callis, Esplin, Haverfield, Rojahn, and Vatta hold stock from Invitae as employees during the conduct of the study. Dr Garcia reports stocks from Invitae as an employee and stocks from 10x Genomics, Personalis, and 23andMe from previous employment outside the submitted work. Dr Nussbaum reports stocks from Invitae as an employee during the conduct of the study; personal fees from Pfizer, Genome Medical, and Maze Therapeutics outside the submitted work; and is a stockholder in Genome Medical and Maze Therapeutics. Dr Rader reports personal fees from Alnylam Pharmaceuticals, Novartis, Pfizer, and Verve as a scientific advisory board member; is chief scientific advisor for Familial Hypercholesterolemia Foundation; and is a cofounder of Vascular Strategies and Staten Biotechnology during the conduct of the study. No other disclosures were reported.

    References
    1.
    Benn  M, Watts  GF, Tybjærg-Hansen  A, Nordestgaard  BG.  Mutations causative of familial hypercholesterolaemia: screening of 98 098 individuals from the Copenhagen General Population Study estimated a prevalence of 1 in 217.   Eur Heart J. 2016;37(17):1384-1394. doi:10.1093/eurheartj/ehw028PubMedGoogle ScholarCrossref
    2.
    Abul-Husn  NS, Manickam  K, Jones  LK,  et al.  Genetic identification of familial hypercholesterolemia within a single U.S. health care system.   Science. 2016;354(6319):aaf7000. doi:10.1126/science.aaf7000PubMedGoogle Scholar
    3.
    Trinder  M, Francis  GA, Brunham  LR.  Association of monogenic vs polygenic hypercholesterolemia with risk of atherosclerotic cardiovascular disease.   JAMA Cardiol. 2020;5(4):390-399. doi:10.1001/jamacardio.2019.5954PubMedGoogle ScholarCrossref
    4.
    Lamiquiz-Moneo  I, Civeira  F, Mateo-Gallego  R,  et al.  Diagnostic yield of sequencing familial hypercholesterolemia genes in individuals with primary hypercholesterolemia.   Rev Esp Cardiol (Engl Ed). Published online July 10, 2020. doi:10.1016/j.rec.2020.06.003PubMedGoogle Scholar
    5.
    Khera  AV, Won  H-H, Peloso  GM,  et al.  Diagnostic yield and clinical utility of sequencing familial hypercholesterolemia genes in patients with severe hypercholesterolemia.   J Am Coll Cardiol. 2016;67(22):2578-2589. doi:10.1016/j.jacc.2016.03.520PubMedGoogle ScholarCrossref
    6.
    Tada  H, Kawashiri  M-A, Nohara  A, Inazu  A, Mabuchi  H, Yamagishi  M.  Impact of clinical signs and genetic diagnosis of familial hypercholesterolaemia on the prevalence of coronary artery disease in patients with severe hypercholesterolaemia.   Eur Heart J. 2017;38(20):1573-1579. doi:10.1093/eurheartj/ehx004PubMedGoogle ScholarCrossref
    7.
    Nordestgaard  BG, Chapman  MJ, Humphries  SE,  et al; European Atherosclerosis Society Consensus Panel.  Familial hypercholesterolaemia is underdiagnosed and undertreated in the general population: guidance for clinicians to prevent coronary heart disease: consensus statement of the European Atherosclerosis Society.   Eur Heart J. 2013;34(45):3478-90a. doi:10.1093/eurheartj/eht273PubMedGoogle ScholarCrossref
    8.
    Versmissen  J, Oosterveer  DM, Yazdanpanah  M,  et al.  Efficacy of statins in familial hypercholesterolaemia: a long term cohort study.   BMJ. 2008;337:a2423. doi:10.1136/bmj.a2423PubMedGoogle ScholarCrossref
    9.
    Luirink  IK, Wiegman  A, Kusters  DM,  et al.  20-Year follow-up of statins in children with familial hypercholesterolemia.   N Engl J Med. 2019;381(16):1547-1556. doi:10.1056/NEJMoa1816454PubMedGoogle ScholarCrossref
    10.
    Fahed  AC, Nemer  GM.  Familial hypercholesterolemia: the lipids or the genes?   Nutr Metab (Lond). 2011;8(1):23. doi:10.1186/1743-7075-8-23PubMedGoogle ScholarCrossref
    11.
    Sturm  AC, Knowles  JW, Gidding  SS,  et al; Convened by the Familial Hypercholesterolemia Foundation.  Clinical genetic testing for familial hypercholesterolemia: JACC Scientific Expert Panel.   J Am Coll Cardiol. 2018;72(6):662-680. doi:10.1016/j.jacc.2018.05.044PubMedGoogle ScholarCrossref
    12.
    Henderson  R, O’Kane  M, McGilligan  V, Watterson  S.  The genetics and screening of familial hypercholesterolaemia.   J Biomed Sci. 2016;23:39. doi:10.1186/s12929-016-0256-1PubMedGoogle ScholarCrossref
    13.
    23andMe. 23andMe personal genome service (PGS) package insert. Published June 2020. Accessed July 24, 2020. https://permalinks.23andme.com/pdf/package_insert_v5.pdf
    14.
    23andMe Blog. 23andMe offers a new report on familial hypercholesterolemia. Published February 20, 2019. Accessed July 24, 2020. https://blog.23andme.com/health-traits/23andme-offers-a-new-report-on-familial-hypercholesterolemia/
    15.
    Lincoln  SE, Kobayashi  Y, Anderson  MJ,  et al.  A systematic comparison of traditional and multigene panel testing for hereditary breast and ovarian cancer genes in more than 1000 patients.   J Mol Diagn. 2015;17(5):533-544. doi:10.1016/j.jmoldx.2015.04.009PubMedGoogle ScholarCrossref
    16.
    Nykamp  K, Anderson  M, Powers  M,  et al; Invitae Clinical Genomics Group.  Sherloc: a comprehensive refinement of the ACMG-AMP variant classification criteria.   Genet Med. 2017;19(10):1105-1117. doi:10.1038/gim.2017.37PubMedGoogle ScholarCrossref
    17.
    Truty  R, Paul  J, Kennemer  M,  et al.  Prevalence and properties of intragenic copy-number variation in mendelian disease genes.   Genet Med. 2019;21(1):114-123. doi:10.1038/s41436-018-0033-5PubMedGoogle ScholarCrossref
    18.
    Richards  S, Aziz  N, Bale  S,  et al; ACMG Laboratory Quality Assurance Committee.  Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology.   Genet Med. 2015;17(5):405-424. doi:10.1038/gim.2015.30PubMedGoogle ScholarCrossref
    19.
    Churchwell  K, Elkind  MSV, Benjamin  RM,  et al; American Heart Association.  Call to action: structural racism as a fundamental driver of health disparities: a presidential advisory from the American Heart Association.   Circulation. 2020;142(24):e454-e468. doi:10.1161/CIR.0000000000000936PubMedGoogle ScholarCrossref
    20.
    eMERGE Clinical Annotation Working Group.  Frequency of genomic secondary findings among 21,915 eMERGE network participants.   Genet Med. 2020;22(9):1470-1477. doi:10.1038/s41436-020-0810-9PubMedGoogle ScholarCrossref
    21.
    Iacocca  MA, Wang  J, Dron  JS,  et al.  Use of next-generation sequencing to detect LDLR gene copy number variation in familial hypercholesterolemia.   J Lipid Res. 2017;58(11):2202-2209. doi:10.1194/jlr.D079301PubMedGoogle ScholarCrossref
    22.
    Regalado  A. More than 26 million people have taken an at-home ancestry test. MIT Technology Review. Published February 11, 2019. Accessed March 5, 2021. https://www.technologyreview.com/2019/02/11/103446/more-than-26-million-people-have-taken-an-at-home-ancestry-test
    23.
    Neal  WA, Knowles  J, Wilemon  K.  Underutilization of cascade screening for familial hypercholesterolemia.   Clin Lipidol. 2014;9(3):291-293. doi:10.2217/clp.14.19Google ScholarCrossref
    24.
    Amor-Salamanca  A, Castillo  S, Gonzalez-Vioque  E,  et al.  Genetically confirmed familial hypercholesterolemia in patients with acute coronary syndrome.   J Am Coll Cardiol. 2017;70(14):1732-1740. doi:10.1016/j.jacc.2017.08.009PubMedGoogle ScholarCrossref
    25.
    Ahmad  ZS, Andersen  RL, Andersen  LH,  et al.  US physician practices for diagnosing familial hypercholesterolemia: data from the CASCADE-FH registry.   J Clin Lipidol. 2016;10(5):1223-1229. doi:10.1016/j.jacl.2016.07.011PubMedGoogle ScholarCrossref
    26.
    Representatives of the Global Familial Hypercholesterolemia Community.  Reducing the clinical and public health burden of familial hypercholesterolemia: a global call to action.   JAMA Cardiol. 2020;5(2):217-229. doi:10.1001/jamacardio.2019.5173Google ScholarCrossref
    27.
    Brown  EE, Sturm  AC, Cuchel  M,  et al.  Genetic testing in dyslipidemia: A scientific statement from the National Lipid Association.   J Clin Lipidol. 2020;14(4):398-413. doi:10.1016/j.jacl.2020.04.011PubMedGoogle ScholarCrossref
    28.
    Musunuru  K, Hershberger  RE, Day  SM,  et al; American Heart Association Council on Genomic and Precision Medicine; Council on Arteriosclerosis, Thrombosis and Vascular Biology; Council on Cardiovascular and Stroke Nursing; and Council on Clinical Cardiology.  Genetic testing for inherited cardiovascular diseases: a scientific statement from the American Heart Association.   Circ Genom Precis Med. 2020;13(4):e000067. doi:10.1161/HCG.0000000000000067PubMedGoogle Scholar
    29.
    Centers for Disease Control and Prevention. More detailed information on key tier 1 applications: familial hypercholesterolemia. Published January 28, 2019. Accessed July 24, 2020. https://www.cdc.gov/genomics/implementation/toolkit/fh_1.htm
    30.
    Taylor  A, Wang  D, Patel  K,  et al.  Mutation detection rate and spectrum in familial hypercholesterolaemia patients in the UK pilot cascade project.   Clin Genet. 2010;77(6):572-580. doi:10.1111/j.1399-0004.2009.01356.xPubMedGoogle ScholarCrossref
    31.
    Haralambos  K, Whatley  SD, Edwards  R,  et al.  Clinical experience of scoring criteria for familial hypercholesterolaemia (FH) genetic testing in Wales.   Atherosclerosis. 2015;240(1):190-196. doi:10.1016/j.atherosclerosis.2015.03.003PubMedGoogle ScholarCrossref
    32.
    Ruhl  GL, Hazel  JW, Clayton  EW, Malin  BA.  Public attitudes toward direct to consumer genetic testing.   AMIA Annu Symp Proc. 2020;2019:774-783PubMedGoogle Scholar
    33.
    Savard  J, Hickerton  C, Metcalfe  SA, Gaff  C, Middleton  A, Newson  AJ.  From expectations to experiences: consumer autonomy and choice in personal genomic testing.   AJOB Empir Bioeth. 2020;11(1):63-76. doi:10.1080/23294515.2019.1701583PubMedGoogle ScholarCrossref
    34.
    Lee  H, Vogel  RI, LeRoy  B, Zierhut  HA.  Adult adoptees and their use of direct-to-consumer genetic testing: Searching for family, searching for health.   J Genet Couns. 2021;30(1):144-157. doi:10.1002/jgc4.1304PubMedGoogle ScholarCrossref
    35.
    US Food and Drug Administration. Direct-to-consumer tests. Updated December 20, 2019. Accessed July 24, 2020. https://www.fda.gov/medical-devices/vitro-diagnostics/direct-consumer-tests
    36.
    Tandy-Connor  S, Guiltinan  J, Krempely  K,  et al.  False-positive results released by direct-to-consumer genetic tests highlight the importance of clinical confirmation testing for appropriate patient care.   Genet Med. 2018;20(12):1515-1521. doi:10.1038/gim.2018.38PubMedGoogle ScholarCrossref
    37.
    Dyrbye  L, Herrin  J, West  CP,  et al.  Association of racial bias with burnout among resident physicians.   JAMA Netw Open. 2019;2(7):e197457. doi:10.1001/jamanetworkopen.2019.7457PubMedGoogle Scholar
    38.
    Sabin  J, Nosek  BA, Greenwald  A, Rivara  FP.  Physicians’ implicit and explicit attitudes about race by MD race, ethnicity, and gender.   J Health Care Poor Underserved. 2009;20(3):896-913. doi:10.1353/hpu.0.0185PubMedGoogle ScholarCrossref
    39.
    Caraballo  C, Massey  D, Mahajan  S,  et al  Racial and ethnic disparities in access to health care among adults in the United States: a 20-year national health interview survey analysis, 1999-2018.   medRxiv. Preprint published online November 4, 2020. doi:10.1101/2020.10.30.20223420Google Scholar
    ×