Genomic Autopsy of Sudden Deaths in Young Individuals

Key Points

Question  Can genomic analysis improve clinical care for families after a sudden death in young individuals?

Findings  In this cohort study of 103 decedents from a national collaboration of medical examiners, pathogenic/likely pathogenic variants in arrhythmia or cardiomyopathy genes were identified in 13%. In multivariate analysis, rare variants in cardiac genes (pathogenic/likely pathogenic/uncertain significance) were associated with younger age at death.

Meaning  Genomic autopsy found clinically actionable cardiac variants in 13% of sudden deaths in young individuals, but a more sophisticated model of multiple gene effects or oligogenic model may be required to optimize clinical care for families.

Importance  Postmortem genetic testing of young individuals with sudden death has previously identified pathogenic gene variants. However, prior studies primarily considered highly penetrant monogenic variants, often without detailed decedent and family clinical information.

Objective  To assess genotype and phenotype risk in a diverse cohort of young decedents with sudden death and their families.

Design, Setting, and Participants  Pathological and whole-genome sequence analysis was conducted in a cohort referred from a national network of medical examiners. Cases were accrued prospectively from May 2015 to March 2019 across 24 US states. Analysis began September 2016 and ended November 2020.

Exposures  Evaluation of autopsy and clinical data integrated with whole-genome sequence data and family member evaluation.

Results  A total of 103 decedents (mean [SD] age at death, 23.7 [11.9] years; age range, 1-44 years), their surviving family members, and 140 sex- and genetic ancestry–matched controls were analyzed. Among 103 decedents, autopsy and clinical data review categorized 36 decedents with postmortem diagnoses, 23 decedents with findings of uncertain significance, and 44 with sudden unexplained death. Pathogenic/likely pathogenic (P/LP) genetic variants in arrhythmia or cardiomyopathy genes were identified in 13 decedents (12.6%). A multivariable analysis including decedent phenotype, ancestry, and sex demonstrated that younger decedents had a higher burden of P/LP variants and select variants of uncertain significance (effect size, −1.64; P = .001). These select, curated variants of uncertain significance in cardiac genes were more common in decedents than controls (83 of 103 decedents [86%] vs 100 of 140 controls [71%]; P = .005), and decedents harbored more rare cardiac variants than controls (2.3 variants per individual vs 1.8 in controls; P = .006). Genetic testing of 31 parent-decedent trios and 14 parent-decedent dyads revealed 8 transmitted P/LP variants and 1 de novo P/LP variant. Incomplete penetrance was present in 6 of 8 parents who transmitted a P/LP variant.

Conclusions and Relevance  Whole-genome sequencing effectively identified P/LP variants in cases of sudden death in young individuals, implicating both arrhythmia and cardiomyopathy genes. Genomic analyses and familial phenotype association suggest potentially additive, oligogenic risk mechanisms for sudden death in this cohort.

Postmortem genetic testing discovers pathogenic or likely pathogenic (P/LP) variants in genes affecting heart rhythm and function among 10% to 25% of people who have sudden unexplained death (SUD) before age 40 years.^1,2 Most studies focused on single genetic variants strongly associated with heart rhythm or cardiac muscle diseases,^3-5 where family screening for P/LP monogenic variants can identify and reduce risk in relatives.^6-8

Many reports considered sudden unknown deaths in which autopsy findings revealed a normal heart.^1,4,9-12 Fewer studies have included hearts with cardiomyopathy phenotypes.^13-16 Therefore, genetic evaluation often centered on arrhythmia genes, with limited phenotyping in the decedent or the family. Additionally, most studies used gene panels, with fewer surveying a broader range of genes (eg, whole-exome analysis) and considering variant inheritance.^17-20

Not every P/LP genetic variant causes a severe phenotype, and this variable penetrance and expressivity should be integrated with clinical evaluation.^21,22 Here, we investigated whether a combined genotype-phenotype approach improved genetic assessment of sudden death. We analyzed the genomes of decedents from a large, ethnically diverse catchment area, with both normal hearts and pathological evidence of cardiomyopathy. We extended the analysis through genetic and clinical evaluations of family members. Finally, we considered a model associating rare variants in a small number of genes with phenotypes (an oligogenic approach).

This work was performed under approval from the institutional review board at Lurie Children’s Hospital. All living human individuals provided informed consent, and next of kin of decedents provided consent. Additional methods are described in detail in the eMethods in Supplement 1.

The Northwestern Sudden Death Collaboration is a prospective consortium of more than 60 county and state coroner and medical examiner offices drawn from 24 US states (eTable 4 in Supplement 1). Coroners and medical examiner offices were recruited from May 2015 to March 2019. Decedents aged between 1 and 45 years at death were included. Cases were excluded if the decedent had been diagnosed with a potentially lethal cardiac disorder prior to sudden death, if coronary artery disease was determined to be the cause of death, or if the autopsy revealed an extracardiac cause of death. Decedents with lethal levels of medications or toxins were excluded (eTable 1 in Supplement 1).

With permission from next of kin, we obtained all available decedent medical records and postmortem evaluations. Decedent phenotypes were reviewed by a multidisciplinary committee of adult and pediatric cardiologists with electrophysiology and cardiomyopathy expertise, geneticists, and a cardiac pathologist. We enrolled as many first-degree relatives as possible.

The control cohort was derived from Northwestern’s biobank, NUgene, where electronic health record data are coupled with a whole-genome sequence, selecting participants older than 40 years without a cardiac diagnosis.²³ We included 140 participants, matched by sex and genetic ancestry to the Northwestern Sudden Death Collaboration decedent cohort.

Whole-genome sequencing was completed and analyzed as detailed in the eMethods in Supplement 1. We constructed a consensus arrhythmia and cardiomyopathy gene panel (cardiac panel; eTable 2 in Supplement 1). Initial selection for rare variants used a gnomAD minor allele frequency less than 0.002, corresponding with a population prevalence of approximately 1 of 500 for clinical cardiomyopathy and channelopathy disease.²⁴ All variants were subsequently confirmed by Sanger sequencing and adjudicated by an independent laboratory based on American College of Medical Genetics (ACMG) criteria. Parental samples were analyzed by site-specific Sanger sequencing.

We performed cellular electrophysiology of KCNQ1 and SCN5A variants not previously studied. These genes were chosen because methods have been established for functional analysis to inform clinical classification according to 2015 ACMG guidelines.²⁵ Variants of unknown significance in KCNQ1 were evaluated functionally using automated whole-cell patch clamp recording of recombinant human K_V7.1 coexpressed with KCNE1 in Chinese hamster ovary (CHO-K1) cells as described previously.²⁶ Variants in SCN5A were evaluated by automated patch clamp recording of recombinant human Na_V1.5 expressed in human embryonic kidney cells (HEK-293T) as described previously.²⁷

Analysis of distribution of variants by age was performed using a t test, and differences among strata were confirmed by nonparametric Wilcoxon rank sum statistics. Analysis of variant of uncertain significance (VUS) proportions was evaluated with χ² test and Wilcoxon rank sum for variants/individual. The VUS heat map was generated in R version 3.6.3 (R Foundation). Variants were normalized by correcting for gene length and multiplying by the gene missense Z score in gnomAD.^28,29 Linear regression was used to evaluate the association of P/LP variants and age at death. The dependence of age at death on the cumulative number of variants present among genes in the cardiac panel was tested using multivariate linear models adjusting for sex and genetic ancestry. Analysis began September 2016 and ended November 2020.

The Northwestern Sudden Death Collaboration prospectively recruited from more than 60 county and state coroner and medical examiner offices (eFigure 1 in Supplement 1). Pathological findings from each decedent were reviewed by a multidisciplinary committee following the workflow outlined in eFigure 2 in Supplement 1. In total, 103 individuals with sudden death were included for analysis (eFigure 3 in Supplement 1). The mean (SD) age at death was 23.7 (11.9) years, and 33 individuals (32%) were 18 years or younger (Table 1). Most decedents were male (77 of 103 [75%]). Activity at time of death included sleep (42 [41%]), awake but at rest (27 [26%]), strenuous activity (15 [15%]), light activity (11 [11%]), and other or not documented (8 [8%]).

Autopsy findings and any available premortem clinical records were used to categorize decedents into 1 of 3 diagnostic categories. A postmortem cardiac diagnosis was present in 36 of 103 decedents (35%). Findings of uncertain significance (FUS) were documented in 23 decedents (22%). The remaining 44 decedents (43%) had no diagnostic findings and were classified as SUD (Figure 1A). The eAppendix in Supplement 1 provides additional phenotype data.

A heart weight exceeding the 95% of normal for height, weight, age, and sex (cardiomegaly)³⁰ was present in 21 of 103 decedents (20%). Cardiomegaly was present in 8 of 10 decedents with hypertrophic cardiomyopathy, 2 of 6 with dilated cardiomyopathy, and 2 of 10 with arrhythmogenic ventricular cardiomyopathy. Isolated cardiomegaly was present in 4 of 23 decedents (17%) with an FUS and in 1 with SUD (1 of 44 [2%]).

We assembled a cardiac panel of 118 genes implicated in inherited cardiomyopathy and arrhythmia conditions. Using this cardiac panel, we identified P/LP variants in 13 of 103 decedents (12.6%). A P/LP variant was present in 6 of 36 decedents (17%) with a postmortem clinical diagnosis, 2 of 23 (9%) with an FUS, and 5 of 44 (11%) classified as SUD. P/LP variants were distributed uniformly across the ages in this cohort (odds ratio, 1.0 [95% CI, 0.96-1.05]; P = .85; Figure 1B). All variants were confirmed by Sanger sequencing and adjudicated as P/LP by an independent clinical genetics laboratory based on ACMG criteria (Table 2).²⁵ Broader genomic analysis beyond the cardiac panel identified 6 additional P/LP variants.

The pathogenic variant FKRP p.Phe56Glyfs*6 was heterozygous in 1 decedent. Based on involvement of FKRP in recessive disease, it was excluded from the P/LP analyses. We similarly excluded the pathogenic familial hypercholesterolemia variant (patient 15; LDLR p.Asp472Tyr) in a decedent younger than 10 years with pathological evidence of arrhythmogenic cardiomyopathy and no evidence of coronary artery disease.

Because variant classification may be influenced by minor allele frequencies in diverse populations, genetic ancestry was determined using principal component analysis of approximately 5 million variants across the genome, creating clusters that correspond with each race and ethnicity. Genetic ancestry and coroner/medical examiner report of race agreed in 87 individuals (84%). Coroner/medical examiner reports did not record Hispanic ancestry, and all instances of mismatch between the coroner/medical examiner’s report of race and genetic ancestry occurred in these cases. The majority of decedents with Hispanic ancestry were reported by coroners or medical examiners as White (9 of 16 [56.3%]) with the remaining decedents reported as Black (1 of 16 [6.3%]) or reported as other by coroners/medical examiners (6 of 16 [37.5%]; eFigure 4 in Supplement 1).

For initial VUS analysis, we considered the 118 genes on the cardiac panel. We identified 292 variants from 80 of these genes. At least 1 P/LP/VUS variant was present in 93 of 103 individuals (90%), with a mean (SD) of 3.0 (1.7) variants per decedent. We identified 205 VUSs among 89 decedents (86%; mean [SD], 2.3 [1.4] VUS per decedent). Almost all variants were heterozygous (204 of 205 [99.5%]). There was 1 hemizygous variant in DMD. For comparison, we evaluated a control cohort, matched for sex and genetic ancestry, without cardiac disease (140 individuals). A smaller percentage of the control population harbored cardiac VUSs (100 of 140 controls [71%] vs 89 of 103 decedents [86%]; P = .005) and there were fewer VUSs per individual in the control cohort (mean [SD], 1.8 [0.9] VUSs per individual in the control group vs 2.3 [1.4] in the decedent group; P = .006).

We expanded this analysis to rare variants throughout the genome (gnomAD global allele frequency <0.002). We sorted for genes expressed in the heart, amino acid changes likely to disrupt function, and clinical pathway databases suggesting genotype-phenotype associations, including those associated with sudden death phenotypes (eMethods in Supplement 1). The high-priority variants were evaluated by the genomic autopsy committee. We identified 41 rare variants of interest in 30 preliminary evidence genes, present within 32 of 103 decedents (31.1%) (eTable 2 in Supplement 1). All rare variants in the preliminary evidence genes were heterozygous except for 1 homozygous splice donor variant in ACSL1 (c.251G>T), present in a teenage decedent with SUD.

Certain genes were more enriched for VUS than would be expected from gene length and population-level variation. Figure 2A is a heat map showing the 35 highest-ranking genes of 79 genes identified with at least 1 curated VUS. Genes illustrated by the darkest colors have the highest weighted number of variants, suggesting excess VUS, and include well-recognized arrhythmia and cardiomyopathy genes such as SCN5A, FLNC, RYR2, and FKRP. In addition, several of the genes on the heat map also include loss-of-function variants that do not meet ACMG criteria for P/LP variants but may have an oligogenic role in disease.

We identified 6 structural variants, 1 each in TSC2, ANK2, CACNA1C, RYR2, HCN2, and TXNRD2 (eTable in Supplement 2). None of these occurred in decedents with another P/LP variant. While these may have contributed to sudden death, limitations of the current classification system kept them from being reliably adjudicated as P/LP/VUS and so they were excluded from classification and variant tabulations.

We obtained DNA from 31 parent-decedent trios and 14 parent-decedent dyads. From these data, 9 P/LP variants were captured; 8 were inherited and 1 was de novo. Clinical data were available from all 8 parents who transmitted a P/LP variant (Table 3). Parents who transmitted a P/LP variant had a range of clinical cardiac findings. Two of 8 transmitting parents had clinical findings concordant with the expected genotype-phenotype association of the decedent’s P/LP variant. One concordant example was a parent with dilated cardiomyopathy without coronary atherosclerosis and who transmitted a pathogenic TTN truncation (patient 8).

The other 6 of 8 gene carriers with P/LP variants demonstrated variable expression or incomplete penetrance. Incomplete penetrance was present in a parent with a normal cardiac phenotype who transmitted a pathogenic PKP2 truncation variant to a decedent child with arrhythmogenic cardiomyopathy (patient 4). Variable expression extended to a parent with normal cardiac imaging and premature ventricular contractions and nonsustained ventricular tachycardia during exercise stress testing who transmitted an MYBPC3 splice variant to a decedent child (patient 7).

Clinically relevant genotype-phenotype mismatch was also observed. For example, a male decedent who died in his early 40s had asymmetric septal hypertrophy (septum, 2.6 cm vs left ventricular free wall, 1.5 cm) and focal perivascular myocyte disarray on histology. Sequencing of this individual identified a pathologic KCNQ1 variant (patient 3; p.Arg594Gln). This variant was inherited from his father, who had aortic valve disease, a normal QT interval, and no ventricular hypertrophy. The decedent’s daughter (in her 20s at the time of clinical evaluation) inherited the variant and had a QTc of 470ms but was asymptomatic.

Absent prior functional reports of KCNQ1-Arg594Gln, we performed in vitro analysis that demonstrated complete loss of function. However, current density recorded in cells coexpressing the abnormal variant plus normal wild-type KCNQ1 to mimic the heterozygous state of the decedent was approximately 70% of the current measured in cells expressing only wild-type KCNQ1 (eFigure 5 in Supplement 1). By contrast, a known dominant-negative KCNQ1 variant (Gly314Ser) exhibited only 20% of wild type current density when coexpressed with wild-type KCNQ1. We conclude KCNQ1-Arg594Gln is a partial loss-of-function variant in the heterozygous context and does not fully explain the phenotype in the parent or offspring.

In contrast, heterozygous KCNQ1 p.Ala300Glu variant was discovered in a female decedent in her second decade of life who exhibited a dominant-negative severe loss of function (eFigure 6 in Supplement 1) consistent with a pathogenic variant causative of type 1 long QT syndrome. These findings are concordant with the patient’s postmortem investigation. The decedent had SUD with normal autopsy, histology, and toxicology evaluation. No premortem cardiac testing had been performed.

Finally, a decedent’s phenotype may be suspicious enough to trigger efforts to reclassify variants of uncertain significance. A teenager died at rest and was found to have cardiomyopathy with cardiomegaly, biatrial and biventricular enlargement, and myocyte hypertrophy (patient 5). Genomic autopsy identified 2 SCN5A variants (p.Thr370Met and Ala691Thr), which were both inherited from the same parent. Both variants were initially classified as VUS, and we performed in vitro experiments using heterologous expression to determine the functional implications of the variants (eFigure 7 in Supplement 1). Whereas p.Ala691Thr (A691T) is a rare population variant (minor allele frequency <0.0001), p.Thr370Met (T370M) has been previously associated with long QT syndrome or sudden death in adults.^31-33 We investigated the functional consequences SCN5A engineered with both T370M/A691T using heterologous expression in HEK293T cells and automated patch clamp recording. Cells expressing SCN5A T370M/A691T exhibited a significantly depolarized voltage dependence of steady-state inactivation and aberrant channel activation during a slow depolarizing voltage ramp consistent with a deleterious gain of function (eFigure 7 and eTable 3 in Supplement 1). Inactivation kinetics were normal and there was no significant difference in persistent current between wild-type and variant channels. When expressed as individual variants in separate experiments, only T370M exhibited these abnormalities (eTable 3 in Supplement 1). In summary, functional analysis of the T370M variant was consistent with a deleterious gain of function and by applying relevant ACMG criteria, this variant was reclassified as LP.²⁵

P/LP genotypes did not fully predict phenotype; therefore, we assessed genetic features across all decedents that may promote risk of death, as manifested by an association between variant burden and age at death. For this analysis, we examined all P/LP/VUS variants in the 118 gene cardiac panel and the 41 rare variants of interest in 30 preliminary evidence genes. We found that having more variants was associated with a younger age at death. For decedents older than 2 years at death, a younger age was associated with a greater number of rare cardiac variants, with an effect size of −1.59 (95% CI, −2.5 to −0.7; P = .001; Figure 2B). A multivariable analysis that included variables for decedent phenotype, ancestry, and sex did not meaningfully change the effect size (−1.64) and the finding remained significant (95% CI, −2.6 to −0.7; P = .001).

This cohort of more than 100 sudden deaths in young individuals, drawn from a diverse population of urban and rural US counties, revealed P/LP variants in cardiac genes in 12.6% of decedents. This yield mirrors a recent multinational postmortem sequencing collaboration,¹ and clinicians can counsel families that several studies with more than 100 decedents have reported that P/LP variants are detectable in 10% to 25% of sudden deaths in young individuals.^2,34,35 P/LP variants were present in 18% of decedents with a postmortem cardiac diagnosis, 9% of decedents with pathological FUS at autopsy, and 11% of those with normal hearts at autopsy (SUD). These results are similar to a 2020 report by Lahrouchi et al¹³ in which evidence of structural cardiac disease was associated with a higher rate of P/LP variants identified postmortem. These data add quantitative support for the recent clinical recommendations to perform a comprehensive postmortem cardiac autopsy and histology in cases of sudden death in young individuals.³⁶

P/LP variants in cardiomyopathy genes were present in decedents with FUS and SUD, indicating cardiomyopathy variants may precede anatomical or histological features and supporting the class 1 recommendation for phenotype-guided clinical screening in first-degree relatives.³⁶ We observed variable and discordant expressivity in first-degree relatives with P/LP variants. Thus, clinical screening should include a broad assessment of cardiac risk without being limited to the decedent’s specific genotype or phenotype.

We identified a quantitative association between risk of sudden death and a multiple-gene model of cardiac disease, in which each variant provides a small contribution to sudden cardiac death risk (an oligogenic approach). We found younger decedents had a higher frequency of rare, potentially deleterious variants, even after correcting for autopsy phenotype, genetic ancestry, and sex, implying that genetic modifiers may influence the age of sudden death. In addition, a higher percentage of decedents had potentially deleterious cardiac VUS when compared with a sex- and ancestry-matched control cohort that had survived past the age of sudden death in young individuals. In current practice, P/LP variants are clinically actionable and VUSs are not. The demonstration of oligogenic effects does not elevate any individual VUS to a clinically actionable status. However, the refinement of oligogenic models may complement the efforts of ClinGen and others to adjudicate highly penetrant cardiorhythm gene variants, leading to improved classification and improved clinical care.³⁷

The diverse catchment area of this study covered US urban and rural areas. Genomic autopsy requires consideration of genetic ancestry to accurately determine what constitutes rare variation within a population. Decedents of Hispanic genetic ancestry were classified as White or other at autopsy, highlighting that medical examiner–derived race and ethnicity may differ from ancestry determined by genomic analysis. Genetic ancestry influences the frequency of VUS, which is potentially important in this population because 86% of decedents had at least 1 VUS (mean, 2.3 VUS per person). Missense VUSs were overrepresented in SCN5A, FLNC, and RYR2, all known cardiomyopathy and arrhythmia genes, suggesting some VUSs may still contribute oligogenic risk.

We found structural variants sudden death genes in 6% of decedents, which may be an underestimate given the difficulty for reliable structural variant detection in short read sequencing.³⁸ We did not attempt to assign pathogenicity to these structural variants given these challenges. This was a voluntary collaboration of more than 60 county and state coroner/medical examiner offices drawn from 24 states that did not capture all sudden deaths in all jurisdictions. Some centers participating in a separate national study chose to refer only individuals 19 years or older,³⁹ so this study cannot accurately estimate an incidence of sudden death in young individuals. Some families declined to participate or only 1 parent was available, limiting full inheritance assessment.

Comprehensive genome sequencing demonstrated P/LP variants in 12.6% of decedents, implicating both arrhythmia and cardiomyopathy genes. P/LP variants were present at similar frequencies in all phenotypes and at all ages from 1 to 45 years. Both genotype and phenotype analyses implicated additional, oligogenic risk factors in sudden death in young individuals.

Corresponding Author: Elizabeth McNally, MD, PhD, Department of Pharmacology, SQ5-516 (elizabeth.mcnally@northwestern.edu), and Alfred L. George Jr, MD, Center for Genetic Medicine, Bluhm Cardiovascular Institute, Searle 820 (al.george@northwestern.edu), Northwestern University Feinberg School of Medicine, 320 E Superior St, Chicago, IL 60611.

Accepted for Publication: June 1, 2021.

Published Online: August 11, 2021. doi:10.1001/jamacardio.2021.2789

Author Contributions: Dr McNally 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. Drs Webster and Puckelwartz contributed equally.

Concept and design: Webster, Puckelwartz, Pesce, Dellefave-Castillo, George, McNally.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Webster, Arunkumar, George, McNally.

Critical revision of the manuscript for important intellectual content: Puckelwartz, Pesce, Dellefave-Castillo, Vanoye, Potet, Page, Kearns, Pottinger, White, Olson, Kofman, Ibrahim, Ing, Brew, Yap, Kadri, George, McNally.

Statistical analysis: Puckelwartz, Pesce, Vanoye, Kearns, Pottinger, McNally.

Obtained funding: Webster, Puckelwartz, George, McNally.

Administrative, technical, or material support: Webster, Vanoye, White, Olson, Kofman, Ibrahim, Brew, Yap, McNally.

Supervision: Potet, George, McNally.

Conflict of Interest Disclosures: Dr Webster reported grants from the National Institutes of Health (NIH)/National Heart, Lung, and Blood Institute and the American Heart Association during the conduct of the study. Dr Puckelwartz reported grants from the NIH/National Heart, Lung, and Blood Institute and the American Heart Association during the conduct of this study. Dr Kofman reported grants from the NIH during the conduct of the study. Dr George reported grants from NIH/National Heart, Lung, and Blood Institute, Praxis Precision Medicines, and Tevard Biosciences and personal fees from consulting for Amgen outside the submitted work. Dr McNally reported grants from the NIH and American Heart Association to Northwestern University during the conduct of the study; personal fees from Amgen, Avidity, AstraZeneca, 4D Molecular Therapeutics, Cytokinetics, Janssen Pharmaceuticals, Pfizer, PepGen, Tenaya, and Invitae outside the submitted work; and is founder of Ikaika Therapeutics. No other disclosures were reported.

Funding/Support: Research reported in this publication was supported, in part, by the National Institutes of Health/National Heart, Lung and Blood Institute (grants KL2TR001424, K23HL130554, R01HL128075, and U01HL131914), the American Heart Association Mentored Clinical and Population Research Award (17MCPRP33660457), the American Heart Association Career Development Award (189CDA34110460), the American Heart Association Strategically Focused Research Network on Sudden Cardiac Death and Arrhythmias (19SFRN34910009), and the Smeds Family Foundation.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: Dr McNally is section editor of JAMA Cardiology but was not involved in any of the decisions regarding review of the manuscript or its acceptance.

Additional Contributions: We gratefully acknowledge the support and participation of all the families who experienced the loss of a loved one. We thank the participating coroner and medical examiner offices. We thank Reshma R. Desai, MS, Tatiana V. Abramova, MS, and Nora Ghabra, MS (Department of Pharmacology, Northwestern University), for technical assistance investigating KCNQ1 and SCN5A variants; these individuals were not compensated outside of their standard salary. We also thank Christina Velázquez, BA, for her family coordination (no compensation was received outside of standard salary) and Jessica Kupsik, BA (no compensation was received outside of standard salary), and Mara Russo, BS (compensation was received), for formatting and graphic contributions (Lurie Children’s Hospital, Division of Cardiology).