Abstract
Purpose
Great uncertainty exists about the costs associated with whole genome sequencing (WGS).
Methods
One hundred cardiology patients with cardiomyopathy diagnoses, and 100 ostensibly healthy primary care patients were randomized to receive a family history report alone or with a WGS report. Cardiology patients also reviewed prior genetic test results. WGS costs were estimated by tracking resource use and staff time. Downstream costs were estimated by identifying services in administrative data, medical records, and patient surveys for 6 months.
Results
The incremental cost per patient of WGS testing was $5,098 in cardiology settings and $5,073 in primary care settings compared to family history alone. Mean six month downstream costs did not differ statistically between the control and WGS arms in either setting (cardiology: difference = −$1,560, 95%CI −$7,558 to $3,866, p=0.36; primary care: difference = $681, 95%CI −$884 to $2,171, p=0.70). Scenario analyses showed the cost reduction of omitting or limiting the types of secondary findings was less than $69 and $182 per patient in cardiology and primary care, respectively.
Conclusion
Short-term costs of WGS were driven by the costs of sequencing and interpretation rather than downstream healthcare. Disclosing additional types of secondary findings has a limited cost impact following disclosure.
Keywords: whole genome sequencing, cardiology, primary care, economics, costs
INTRODUCTION
Whole genome sequencing (WGS) can facilitate molecular diagnoses and identify genetic variants to characterize disease risks, tailor medications, screen for recessive traits and more. Dramatic improvements in its cost, speed, and capabilities are fueling expectations that WGS will become an important part of everyday patient care,1,2 and some commentators hope that it will streamline diagnoses and enhance disease prevention.3 Early evidence is promising,4,5 but there are concerns that WGS may also initiate a cascade of confirmatory testing and ongoing screening that greatly increases healthcare expenditures.6,7 Empirical data to inform the discussion are sparse.
Studies to date have examined the costs of genomic sequencing to provide molecular diagnoses for specific syndromes.8–13 or have projected the potential cost impact of disclosing a limited set of secondary findings.14 However, these early attempts do not fully account for more extensive secondary findings such as pharmacogenomic applications and risk predictions for common disease. Furthermore, current modeling efforts assume rational responses despite evidence that patients often respond to secondary findings in unexpected ways15,16 and evidence that physicians are often unclear about their obligations to act on information that is unrelated to the primary indication for testing.17 Few clinical studies have examined the costs of integrating WGS into the everyday care of patients, including broader applications that capitalize on the full potential of WGS.
To fill these knowledge gaps, we conducted a cost analysis of the MedSeq Project, a pair of randomized controlled trials of WGS in cardiology and primary care settings. Limited prior analyses within the primary care setting revealed some evidence for increased healthcare utilization and costs within the Partners HealthCare system following WGS.18 The analyses presented here extend these findings through a more comprehensive accounting of in- and out-of-system healthcare services and out-of-pocket costs in both cardiology and primary care settings, including a micro-costing analysis of WGS itself and scenario and sensitivity analyses that examine the impact of different reporting strategies and cost assumptions. The objective of these analyses is to provide novel insight about the short-term cost impact of integrating WGS into medical care.
METHODS
Clinical Trial Overview
The rationale, design and primary molecular and clinical findings of the MedSeq Project were previously reported,17–23 and methodological details are summarized in detail in Appendix 1. Briefly, primary care physicians identified ostensibly healthy patients between 40 and 65 years of age and cardiologists identified adult patients of any age with diagnoses of hypertrophic or dilated cardiomyopathy (HCM/DCM) for recruitment by MedSeq Project staff. To compare panel-based genetic testing to WGS in patients with HCM/DCM, cardiology patients were either confirmed to have had panel testing, or provided panel testing prior to MedSeq Project disclosure sessions. At their first study visit, patients consented for participation, provided family history information, had blood drawn, and provided baseline information. Using concealed envelopes, study staff then randomly assigned patients to a “control arm” that received a structured family history review (FH) or a “WGS arm” that received FH plus a WGS analysis and report. Cardiologists also reviewed their patients’ prior genetic test results in both randomization arms.
WGS reports included findings related to the patients’ condition (cardiology patients only), monogenic disease risks associated with well-established genetic conditions, carrier status for autosomal recessive conditions, pharmacogenomic findings, and risk predictions based on markers for cardiometabolic traits.19,20,22 Prior to disclosure, physicians could contact a Genome Resource Center staffed by genetic counselors and medical geneticists to ask questions about reports or study procedures. Investigators were not blinded to randomization, although patient participants were blinded until disclosure sessions and physician participants were blinded until they received family history and WGS reports, if applicable. The Partners Human Research Committee and Baylor College of Medicine Institutional Review Board both approved the study protocol (ClinicalTrials.gov #NCT01736566). The protocol, statistical code, and data set are available upon request from the corresponding author. Genetic data is available via dbGaP (accession #phs000958.v1.p1).
Cost-Analysis
Overall Approach
We undertook a planned cost analysis alongside the pilot randomized controlled trial using guidelines published by the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) and the Second Panel on Cost-Effectiveness in Health and Medicine.24,25 Costs are presented in 2015 US dollars.
Costs of WGS
To understand the costs of WGS to the healthcare system, we used a microcosting approach with program operation logs,26 where staff measured personnel effort and resource use using a combination of forms that tracked each DNA sample and by reviewing audiorecorded consent and disclosure sessions to determine their lengths. Consent sessions for MedSeq Project participation were assumed to be comparable in length to consent sessions for WGS. Personnel effort was translated to costs using 2015 wage data as described in more detail in Appendix 4. We assigned costs for disclosure sessions based on Centers for Medicare and Medicaid Services (CMS) fee schedules for outpatient evaluation and management visits. The cost of sequencing was estimated from data provided by the National Human Genome Research Institute,27 while costs for confirming monogenic and carrier findings via Sanger sequencing were based on 2015 Laboratory for Molecular Medicine rates for familial variant testing.
Downstream Healthcare Utilization and Costs
To understand the short-term impact of WGS disclosure on follow-up health sector spending, we assessed costs over the six months preceding results disclosure and the six months following disclosure for both randomization arms. We compared post-disclosure costs by randomization status, as well as separately comparing whether the difference in pre-disclosure and post-disclosure costs differed by randomization status. We supplemented healthcare services identified in administrative data accessed via the Partners HealthCare Research Patient Data Registry (RPDR) with services identified through a review of medical records. We also incorporated services reported in patient surveys completed at disclosure sessions and 6 weeks and 6 months post-disclosure that were not identified in RPDR or medical records data. Costs for services were estimated by multiplying utilization by cost weights derived from CMS fee schedules as described in Appendix 4. Patient out-of-pocket expenses during the post-disclosure period were assessed using survey items adapted from the McMaster Cost of Care Questionnaire,28 where patient participants reported average monthly costs for procedural and visit co-pays, medical equipment, medications, and additional expenses related to their healthcare. Differences from prior-reported costs18 are primarily due to the addition of services identified through patient self-report.
Sensitivity and Scenario Analyses
Sensitivity and scenario analyses incorporating both WGS costs and downstream costs examined how findings varied with different analytic assumptions. First, we examined how costs changed when different types of downstream healthcare services were included. One approach examined “immediately attributable” services, including only services that physicians reported recommending in response to FH review and/or FH review and WGS reports, and that were confirmed through medical record review or patient self-report. Another approach included estimates of the costs for a comprehensive work-up of monogenic disease risks per guidelines such as GeneReviews, clinical synopses such as Online Mendelian Inheritance in Man, and published medical literature.29,30 We also examined the implications of different reporting criteria, such as omitting carrier status or polygenic risk predictions, and reporting no secondary findings or reporting only secondary monogenic disease risk and carrier findings classified as likely pathogenic or pathogenic, per ACMG recommendations.31 In addition, we varied cost assumptions of WGS to reflect the range of prices available at CLIA-certified labs, and we varied healthcare price weights to be 50% to 200% of CMS fee schedule amounts. Finally, we analyzed costs from a third-party payer perspective by omitting patient out-of-pocket expenses.
Statistical Approach
A priori enrollment goals were not established given that this was a pilot study. We used multivariable linear regression to compare arithmetic mean costs by randomization arms, as recommended,24 including all observed data and imputing data for missing observations. Models included terms for WGS vs control randomization status, cardiology vs primary care, an interaction term, and baseline health status. Covariates were included if they improved model precision (i.e., p<0.05). We also used this approach to compare the difference in health care spending in the six month period post-disclosure against the six month period pre-disclosure, but omitted patient out-of-pocket expenses because relevant survey data were not available. Sub-analyses included additional terms as relevant to examine changes over time, to examine the impact of learning about an unexpected monogenic disease risk or to examine the impact of cardiometabolic risk predictions. Although terms for interactions between cohorts and randomization status were not significant, we report cohorts separately because downstream costs among cardiology patients were much larger than downstream costs among primary care patients. To compare changes in health care utilization, we used nonparametric tests (Wilcoxon rank-sum and Wilcoxon signed-rank tests) because data were highly skewed.
Four randomized individuals who did not attend disclosure sessions were omitted from analyses (Appendix 5). Analyses were conducted using R version 3.4.3. As recommended for smaller samples, confidence intervals were calculated using nonparametric bootstrapping (1,000 samples).24 We assumed that missing survey data were missing at random, and imputed non-response with fully conditional specification, running 20 iterations to create each of 20 imputed datasets for each bootstrapped sample. Imputed data accounted for less than 1% of WGS costs through disclosure and 1.2% of downstream medical costs.
RESULTS
Sample Characteristics and Whole Genome Sequencing Findings
Two hundred participants attended results disclosure sessions. Demographic characteristics did not vary by randomization status except on educational attainment among primary care patients (Table 1). Table 2 summarizes monogenic and carrier findings among patients who received WGS. All cardiomyopathy variants that had been identified during panel testing of cardiology patients were confirmed by WGS, although an 18-base pair duplication in MYBPC3 may have been missed if investigators did not know to look for it because it had been identified during prior targeted HCM genetic testing. In addition, among the 22 cardiology patients randomized to WGS who had no variants identified in panel testing, a pathogenic variant associated with HCM was identified in one. Among the 27 cardiology patients who had positive or inconclusive panel testing results, three with inconclusive prior results were identified with a previously-unreported variant of uncertain significance associated with their DCM or HCM diagnosis. Secondary findings about monogenic disease risks and carrier status for autosomal recessive conditions were identified in 8 (16%) and 41 (84%) cardiology patients in the WGS arm, respectively.
Table 1.
Cardiology | Primary Care | |||
---|---|---|---|---|
Characteristic n (%) unless noted |
Control (n=51) |
WGS (n=49) |
Control (n=50) |
WGS (n=50) |
Mean age (sd) | 55.6 (12.2) | 56.1 (16.2) | 54.6 (7.6) | 55.1 (7.1) |
Age range | 26.0 – 72.1 | 18.7 – 84.6 | 41.6 – 67.9 | 41.2 – 65.9 |
Gender | ||||
Female | 19 (37.3%) | 24 (49.0%) | 30 (60.0%) | 28 (56.0%) |
Male | 32 (62.7%) | 25 (51.0%) | 20 (40.0%) | 22 (44.0%) |
Race | ||||
Non-Hispanic white | 45 (88.2%) | 43 (87.8%) | 43 (86.0%) | 44 (88.0%) |
Other | 6 (11.8%) | 6 (12.2%) | 7 (14.0%) | 6 (12.0%) |
Annual Household Income | ||||
<$100,000 | 19 (37.3%) | 24 (49.0%) | 16 (32.0%) | 9 (18.0%) |
≥$100,000 | 30 (58.8%) | 23 (46.9%) | 31 (62.0%) | 40 (80.0%) |
No response | 2 (3.9%) | 2 (4.1%) | 3 (6.0%) | 1 (2.0%) |
Education | ||||
Did not graduate from college | 9 (17.6%) | 13 (26.5%) | 11 (22.0%) | 3 (6.0%) |
College graduate or higher | 42 (82.4%) | 36 (73.5%) | 39 (78.0%)* | 47 (94.0%)* |
Diagnosis (Cardiology Cohort) | ||||
Hypertrophic cardiomyopathy | 37 (72.5%) | 42 (85.7%) | ||
Dilated cardiomyopathy | 14 (27.5%) | 7 (14.3%) | ||
Has health insurance | 51 (100%) | 49 (100%) | 49 (98.0%) | 49 (98.0%) |
Self-reported health | ||||
Excellent | 4 (8%) | 2 (4%) | 16 (32%) | 23 (46%) |
Very good | 19 (37%) | 16 (33%) | 24 (48%) | 21 (42%) |
Good | 20 (39%) | 20 (41%) | 8 (16%) | 4 (8%) |
Fair | 6 (12%) | 9 (18%) | 2 (4%) | 2 (4%) |
Poor | 2 (4%) | 2 (4%) | 0 (0%) | 0 (0%) |
p=0.021 in within-cohort analyses
Table 2.
Variants Related to Condition |
Monogenic Disease Risk Secondary Findings |
Carrier Findings | |||||||
---|---|---|---|---|---|---|---|---|---|
Variant Classification | Patients with this result (%) |
Mean per patient (sd) |
Range per patient |
Patients with this result (%) |
Mean per patient (sd) |
Range per patient |
Patients with this result (%) |
Mean per patient (sd) |
Range per patient |
Cardiology cohort (n=49) | |||||||||
Any results | 24 (49%) | 0.6 (0.6) | 0–2 | 8 (16%) | 0.2 (0.4) | 0–1 | 41 (84%) | 1.9 (1.4) | 0–6 |
Pathogenic | 12 (24%) | 0.2 (0.4) | 0–1 | 2 (4%) | 0.0 (0.2) | 0–1 | 33 (67%) | 1.2 (1.1) | 0–4 |
Likely pathogenic | 4 (8%) | 0.1 (0.3) | 0–1 | 3 (6%) | 0.1 (0.2) | 0–1 | 18 (37%) | 0.4 (0.6) | 0–2 |
VUS: Favor pathogenic | 4 (8%) | 0.1 (0.3) | 0–1 | 1 (2%) | 0.0 (0.1) | 0–1 | 6 (12%) | 0.2 (0.6) | 0–3 |
VUS* | 5 (10%) | 0.1 (0.5) | 0–2 | - | - | - | - | - | - |
Risk allele† | 0 (0%) | 0.0 (0.0) | 0 | 2 (4%) | 0.0 (0.2) | 0–1 | 1 (2%) | 0.0 (0.1) | 0–1 |
Primary care cohort (n=50) | Monogenic Disease Risk Findings | Carrier Findings | |||||||
All results | 13 (26%) | 0.3 (0.4) | 0–1 | 50 (100%) | 2.7 (1.5) | 1–7 | |||
Pathogenic | 6 (12%) | 0.1 (0.3) | 0–1 | 45 (90%) | 1.9 (1.3) | 0–5 | |||
Likely pathogenic | 3 (6%) | 0.1 (0.2) | 0–1 | 21 (42%) | 0.6 (0.8) | 0–3 | |||
VUS: Favor pathogenic | 3 (6%) | 0.1 (0.2) | 0–1 | 10 (20%) | 0.2 (0.5) | 0–2 | |||
Risk allele† | 1 (3%) | 0.0 (0.1) | 0–1 | 0 (0%) | 0 (0.0) | 0 |
Variants of uncertain significance (VUS) where evidence did not favor pathogenicity were only disclosed if they potentially explained a cardiology patient’s diagnosis of hypertrophic or dilated cardiomyopathy. All patients also received polygenic risk predictions about 8 cardiometabolic traits and pharmacogenomic results about 5 drugs. See Appendix 6 and Appendix 7, respectively
Two cardiology patients and one primary care patient were identified with risk alleles for Factor V Leiden thrombophilia, which were reported as monogenic disease risk findings. In addition, one cardiology patient was identified with an allele associated with HEXA pseudodeficiency which was reported as a carrier finding.
Among the 50 primary care patients who received WGS, we identified variants associated with monogenic disease in 13 (26%), although two patients who received information about hereditary hemochromatosis had already been diagnosed with the condition. All primary care patients in the WGS arm (100%) received information about carrier status for autosomal conditions. Specific variants are detailed in Appendix 6 (monogenic disease risk findings) and Appendix 7 (carrier findings).
As noted previously, all cardiology and primary care patients in the WGS arms also received polygenic risk predictions for eight cardiometabolic traits and pharmacogenomic information for five drugs, summarized in Appendix 8.
Cost of Sequencing
Table 3 summarizes the time demands and associated costs of family history review and WGS reporting. The average incremental cost per patient for WGS, including variant interpretation and disclosure, was $5,098 in cardiology settings and $5,073 in primary care settings, with the large majority attributed to sequencing itself ($4,000). Across cohorts, the study incurred an additional $613 (12% of total WGS costs) per patient randomized to WGS, on average, to confirm monogenic and carrier status variants via Sanger sequencing; and subanalyses of patients randomized to WGS showed an average incremental cost of $111 for each variant reported (p<0.001). Laboratory personnel time for variant interpretation and reporting increased by 13% for every variant reported (p<0.001), while overall time demands decreased by an average of 158 minutes per patient per study year (p=0.033).
Table 3.
Mean Time (SD) | Mean Costs, USD (SD) | Difference in Mean Costs, USD (95% CI) |
p | |||
---|---|---|---|---|---|---|
Cardiology Cohort | Control (n=51) | WGS (n=49) | Control (n=51) | WGS (n=49) | ||
Informed consent for WGS | 0 (0) | 16 (6) | 0 (0) | 15 (5) | 15 (13 to 17) | <0.001 |
Sequencing† | - | - | 0 (0) | 4000 (0) | 4000 (4000 to 4000) | <0.001 |
Sanger confirmation† | - | - | 0 (0) | 603 (118) | 603 (571 to 637) | <0.001 |
Variant interpretation and report drafting | 0 (0) | 305 (219) | 0 (0) | 292 (209) | 292 (231 to 356) | <0.001 |
Laboratory report signout | 0 (0) | 82 (54) | 0 (0) | 101 (66) | 101 (83 to 120) | <0.001 |
Data storage† | - | - | 0 (0) | 75 (0) | 75 (75 to 75) | <0.001 |
Consultation with Genome Resource Center | 0 (0) | 0 (2) | 0 (0) | 0 (2) | 0 (0 to 1) | 0.863 |
Disclosure of results | 14 (9) | 24 (11) | 170 (27) | 183 (25) | 12 (2 to 22) | 0.026 |
Total, cardiology | 14 (9) | 428 (228) | 170 (27) | 5268 (288) | 5098 (5022 to 5182) | <0.001 |
Primary Care Cohort | Control (n=50) | WGS (n=50) | Control (n=50) | WGS (n=50) | ||
Informed consent for WGS | 0 (0) | 16 (5) | 0 (0) | 14 (5) | 14 (13 to 16) | <0.001 |
Sequencing† | - | - | 0 (0) | 4000 (0) | 4000 (4000 to 4000) | <0.001 |
Sanger confirmation† | - | - | 0 (0) | 624 (120) | 624 (591 to 660) | <0.001 |
Variant interpretation and report drafting | 0 (0) | 214 (155) | 0 (0) | 204 (148) | 204 (162 to 248) | <0.001 |
Laboratory report signout | 0 (0) | 95 (63) | 0 (0) | 116 (77) | 116 (95 to 139) | <0.001 |
Data storage† | 0 (0) | 0 (0) | 0 (0) | 75 (0) | 75 (75 to 75) | <0.001 |
Consultation with Genome Resource Center | 0 (1) | 9 (23) | 0 (1) | 6 (17) | 6 (2 to 11) | 0.003 |
Disclosure of results | 17 (10) | 35 (18) | 149 (26) | 183 (28) | 34 (23 to 44) | <0.001 |
Total, primary care | 17 (10) | 368 (184) | 149 (27) | 5222 (273) | 5073 (4995 to 5155) | <0.001 |
Cardiology patients also received a review of findings from prior genetic testing for their cardiomyopathy diagnoses. Analyses exclude research-specific costs, such as consenting participants for study participation. Except where noted, costs were estimated using a microcosting approach by tracking time demands and applying relevant wage rates. Steps that were not applicable to participants randomized to the control arm were not assigned times and assigned zero for associated costs.
Costs for these steps were estimated from market rates, and we did not track associated laboratory time demands.
Notably, cardiologists consulted the Genome Resource Center only once during the study, to discuss carrier status findings (compared to 11 consultations in primary care with a variety of questions). In addition, WGS increased the length of cardiology disclosure sessions by an average of 10.7 minutes (WGS arm: 15.3 min; control arm: 4.6 min; p<0.001) and primary care disclosure sessions by an average of 18.7 min (WGS arm: 30.1 min; control arm: 11.4 min; p<0.001). The added length of WGS disclosure sessions increased costs $23 per patient, on average.
Downstream Healthcare Utilization and Costs
Healthcare utilization and costs from disclosure through six months post-disclosure from a healthcare sector perspective are summarized in Table 4, and commonly-observed procedures during the post-disclosure period are summarized in Appendix 9. In cardiology care, mean six month total downstream costs were $1,560 lower in the WGS arm compared to the control arm (95%CI: −$7,558 to $3,866, p=0.357), although mean downstream costs in the WGS arm were $700 higher when hospitalizations were omitted (95%CI: −$1,634 to $3,440, p=0.393). In primary care, mean six month total downstream costs were $681 greater in the WGS arm compared to the control arm (95%CI: −$884 to $2,171, p=0.699) when hospitalizations were included and $374 greater (95%CI: −$991 to $1,570) when hospitalizations were omitted.
Table 4.
Mean Counts (SD) |
Mean Cost, USD |
Difference in Mean Costs, USD (95% CI) |
p | |||
---|---|---|---|---|---|---|
Cardiology Cohort | Control (n=51) | WGS (n=49) | Control (n=51) | WGS (n=49) | ||
Visits | 7.2 (7.5) | 7.8 (6.5) | 1347 | 1344 | −3 (−517 to 488) | 0.504 |
Labs, outpatient care | 6.9 (6.8) | 9.5 (15.5) | 109 | 137 | 28 (−39 to 132) | 0.388 |
Imaging tests, outpatient care | 2.1 (1.9) | 2.1 (2.2) | 638 | 696 | 58 (−300 to 445) | 0.385 |
Cardiology tests, outpatient care | 3.2 (1.7) | 3.1 (1.9) | 188 | 192 | 4 (−42 to 52) | 0.473 |
Other outpatient procedures | † | † | 1489 | 2265 | 776 (−1401 to 3091) | 0.337 |
Hospitalizations | 0.3 (0.8) | 0.2 (0.4) | 4978 | 2717 | −2261 (−7283 to 2459) | 0.280 |
Medications | 4.5 (2.6) | 5.2 (3.1) | 437 | 468 | 31 (−211 to 292) | 0.389 |
Medical equipment | † | † | 39 | 20 | −19 (−65 to 17) | 0.318 |
Visit co-pays | † | † | 319 | 208 | −111 (−409 to 126) | 0.516 |
Patient procedural co-pays | † | † | 75 | 46 | −29 (−84 to 23) | 0.627 |
Additional out-of-pocket | † | † | 51 | 15 | −36 (−94 to 7) | 0.406 |
Total, Cardiology | 9670 | 8109 | −1560 (−7558 to 3866) | 0.357 | ||
Excluding hospitalizations | 4692 | 5392 | 700 (−1634 to 3440) | 0.393 | ||
Primary Care Cohort | Control (n=50) | WGS (n=50) | Control (n=50) | WGS (n=50) | ||
Visits | 6.9 (7.4) | 8.4 (8.7) | 1347 | 1581 | 234 (−317 to 759) | 0.403 |
Labs, outpatient care | 4.4 (4.8) | 5.5 (5.9) | 66 | 96 | 30 (−9 to 76) | 0.410 |
Imaging tests, outpatient care | 0.9 (1.5) | 1.0 (1.4) | 149 | 227 | 78 (−40 to 205) | 0.586 |
Cardiology tests, outpatient care | 0.4 (0.6) | 0.5 (0.9) | 49 | 52 | 3 (−28 to 39) | 0.579 |
Other outpatient procedures | † | † | 540 | 625 | 85 (−460 to 634) | 0.787 |
Hospitalizations | 0.0 (0.2) | 0.0 (0.1) | 142 | 449 | 307 (−473 to 1211) | 0.811 |
Medications | 2.0 (2.1) | 1.8 (1.9) | 159 | 157 | −2 (−108 to 99) | 0.674 |
Medical equipment | † | † | 7 | 4 | −3 (−24 to 12) | 0.645 |
Visit co-pays | † | † | 357 | 317 | −41 (−518 to 287) | 0.424 |
Patient procedural co-pays | † | † | 110 | 150 | 40 (−124 to 231) | 0.370 |
Additional out-of-pocket | † | † | 62 | 11 | −52 (−183 to 17) | 0.371 |
Total, Primary Care | 2989 | 3670 | 681 (−884 to 2171) | 0.699 | ||
Excluding hospitalizations | 2847 | 3221 | 374 (−991 to 1570) | 0.577 |
Cardiology patients also received a review of findings from prior genetic testing for their cardiomyopathy diagnoses.
Counts data were not collected for these items.
Analyses that compared healthcare utilization and costs in the post-disclosure period to the pre-disclosure period showed some increases. Among cardiology participants (Appendix 10), we observed more imaging tests (all p≤0.047) in both arms in the six months following disclosure compared to the six months preceding disclosure, although changes over time on costs were observed only for visits (all p≤0.002) and for imaging tests in the WGS arm (p=0.048). Changes over time in total costs were non-significant, with an increase of $2,939 (95%CI: −$2,857 to $8,772) in the control arm and an increase of $1,728 (95%CI: −$3,067 to $6,387) in the WGS arm (Appendix 11). Changes over time did not differ by randomization status (p=0.376).
Among primary care participants and comparing the post-period to the pre-disclosure period, we observed more healthcare visits in the WGS arm (p=0.020) and greater visit-related costs in both randomization arms (all p<0.001). Changes in total costs were non-significant, with an increase of $682 (95%CI: −$472 to $1,680) in the control arm and an increase of $1,919 (95%CI: −$1,151 to $2,828) in the WGS arm. Changes over time did not differ by randomization status (p=0.572).
Sub-analyses of participants receiving WGS showed no statistically significant association between mean downstream costs among individuals with a previously unknown monogenic disease risk and individuals without one in either primary care ($4,800 vs. $2,900, respectively, p=0.76) or cardiology ($6,861 vs $9,611, respectively, p=0.47). Analyses on specific categories of costs were also inconclusive (all p≥0.161).
Sensitivity and Scenario Analyses
Sensitivity analyses (Table 5) compared randomization arms on total costs, including WGS and downstream costs. Varying the WGS costs had the largest impact, where differences between randomization arms were observed at $10,000 but not at $500. Among cardiology patients, differences between randomization arms were observed when the WGS costs were $10,000, when healthcare services were assigned cost weights at 50% of CMS fee schedule amounts, and when only immediately attributable services were analyzed. Notably, immediately attributable services (Appendix 12) were observed for only one cardiology patient in the control arm and one cardiology patient in the WGS arm. Among primary care patients, differences between randomization arms were also observed in all analyses except when WGS costs were $2,500 or less and when healthcare services were assigned cost weights at 150% or more of CMS fee schedule amounts (Appendix 13). Among cardiology patients, omitting secondary findings altogether reduced costs by $69 per patient, whereas among primary care patients, limiting reporting to only pathogenic variants for monogenic disease and carrier status reduced costs by $182 per patient.
Table 5.
Costs Considered and Parameter Assumptions |
Control Arm, USD |
WGS Arm, USD |
Difference between Arms, USD (95% CI) |
p |
---|---|---|---|---|
Cardiology Cohort | (n = 51) | (n = 49) | ||
Base case | 9841 | 13376 | 3535 (−2490 to 8970) | 0.236 |
Downstream costs analyzed | ||||
Immediately attributable services | 176 | 5270 | 5094 (5015 to 5175) | <0.001 |
Full genetics workup | 9841 | 13400 | 3559 (−2457 to 8980) | 0.234 |
Secondary findings reported | ||||
Omitting carrier status | 9841 | 13194 | 3353 (−2634 to 8780) | 0.247 |
Omitting polygenic risk predictions | 9841 | 13372 | 3531 (−2494 to 8971) | 0.236 |
Reporting criteria for secondary findings about monogenic disease risks and carrier status | ||||
Pathogenic or likely pathogenic | 9841 | 13250 | 3409 (−2595 to 8838) | 0.243 |
Only pathogenic | 9841 | 13244 | 3403 (−2601 to 8839) | 0.244 |
No secondary findings | 9941 | 13307 | 3366 (−1042 to 7774) | 0.134 |
Total cost of integrating WGS (base: $5,268) | ||||
$500 | 9841 | 8610 | −1231 (−7234 to 4198) | 0.371 |
$1000 | 9841 | 9110 | −731 (−6734 to 4698) | 0.387 |
$2500 | 9841 | 10609 | 769 (−5234 to 6198) | 0.374 |
$10000 | 9841 | 18110 | 8269 (2266 to 13698) | 0.035 |
Healthcare costs | ||||
50% of CMS rates | 5838 | 10058 | 4219 (1059 to 7044) | 0.038 |
75% of CMS rates | 7840 | 11717 | 3877 (−712 to 7983) | 0.144 |
150% of CMS rates | 13845 | 16695 | 2850 (−6094 to 11029) | 0.323 |
200% of CMS rates | 17848 | 20013 | 2165 (−9718 to 13041) | 0.358 |
Perspective | ||||
Health sector (base case) | 9841 | 13376 | 3535 (−2490 to 8970) | 0.236 |
Third-party payer (excludes out-of-pocket costs) | 9397 | 13107 | 3710 (−2272 to 9082) | 0.220 |
Primary Care Cohort | (n = 50) | (n = 50) | ||
Base case | 3137 | 8894 | 5756 (4196 to 7232) | 0.017 |
Downstream costs analyzed | ||||
Immediately attributable services | 196 | 5290 | 5094 (4982 to 5211) | <0.001 |
Full genetics workup | 3137 | 9018 | 5881 (4330 to 7388) | 0.015 |
Secondary findings reported† | ||||
Omitting carrier status | 3137 | 8668 | 5531 (3974 to 7014) | 0.021 |
Omitting polygenic risk predictions | 3137 | 8878 | 5741 (4177 to 7224) | 0.017 |
Reporting criteria for monogenic disease risk and carrier status findings | ||||
Pathogenic or likely pathogenic | 3137 | 8772 | 5635 (4079 to 7124) | 0.019 |
Only pathogenic | 3137 | 8712 | 5576 (4031 to 7029) | 0.020 |
Total cost of integrating WGS (base: $5,136) | ||||
$500 | 3137 | 4169 | 1032 (−538 to 2525) | 0.617 |
$1000 | 3137 | 4669 | 1532 (−38 to 3025) | 0.490 |
$2500 | 3137 | 6169 | 3032 (1462 to 4525) | 0.186 |
$10000 | 3137 | 13669 | 10532 (8962 to 12025) | <0.001 |
Healthcare costs | ||||
50% of CMS rates | 2331 | 7792 | 5461 (4331 to 6480) | <0.001 |
75% of CMS rates | 2734 | 8343 | 5609 (4279 to 6866) | 0.003 |
150% of CMS rates | 3942 | 9996 | 6054 (3946 to 8125) | 0.077 |
200% of CMS rates | 4748 | 11098 | 6350 (3689 to 9031) | 0.154 |
Perspective | ||||
Health sector (base case) | 3137 | 8894 | 5756 (4196 to 7232) | 0.017 |
Third-party payer (excludes out-of-pocket costs) | 2607 | 8416 | 5809 (4571 to 7091) | 0.013 |
Analyses compared randomization arms after varying the components that were examined or cost assumptions. Costs included those associated with reviewing family history reports, prior genetic test results (cardiology cohort, only) and whole genome sequencing reports, if applicable; and additional health-related costs over the six-month period afterwards.
Scenario analyses of primary care patients assumed monogenic disease risks were considered primary findings.
DISCUSSION
As the first randomized trial of WGS in primary and specialty care, the MedSeq Project presents a unique opportunity to collect novel data about the costs of WGS. Our data showed that the short-term costs were primarily driven by the costs of sequencing, interpretation, and disclosure, and we did not find evidence that WGS increased downstream healthcare costs. Although the MedSeq Project was not powered to draw definitive conclusions about the cost impact of WGS in cardiology and primary care settings, these findings may provide an important early foundation for future trials.
The patient-level data from the MedSeq Project adds to the paucity of economic literature surrounding WGS, particularly patient level data.32 Our findings are in line with previous work that provided microcosting data about whole exome sequencing,9 and extend them by providing insight about whole genome sequencing and how costs may vary with different reporting strategies and cost assumptions. Advances in next generation sequencing approaches are continuing to make WGS less expensive,27 but approximately 25% of the total costs for providing WGS in our study were for steps such as variant confirmation and bioinformatic analyses. Nevertheless, it is possible that the costs associated with these associated steps will also drop, given recommendations to forego confirmation of nucleotide substitutions via Sanger sequencing.33 In addition, the emergence of large genomic data sets such as ExAC,34 improvements in data sharing between laboratories, and innovations in approaches to variant interpretation such as cross-species analyses35 are likely to decrease bioinformatic costs even further by reducing the number of variants that require manual review and classification. Open questions also remain about whether the incremental benefits of WGS are worth its incremental costs compared to approaches such as exome sequencing, which is typically at least 25% cheaper expensive, or gene panels which can be thousands of dollars less expensive.
Notably, our study showed that WGS had a limited impact on the actions of cardiologists. Cardiologists in our study consulted the Genome Resource Center only once for assistance, spent only 10 additional minutes during disclosure sessions addressing WGS reports, and ordered immediate follow-up testing in response to WGS findings in only one instance. Cardiologists in our study tended to be familiar with genetic testing and regularly reviewed their patients’ family histories, and all of the cardiology patients had received panel genetic testing for their diagnoses prior to disclosure sessions. It is possible that cardiologists had already addressed any important family history patterns, and that they believed they already had most of the information they needed to judge the importance of WGS findings, despite the novelty of most secondary findings.
While the short-term impact on patient wellbeing was not assessed, WGS did identify health-relevant variants in a large majority of participants, including diagnosis-related variants in half of sequenced cardiology patients and additional monogenic disease risks in eight cardiology patients and 13 primary care patients. The clinical significance of most of these variants was unclear. Guidelines to help physicians respond to monogenic findings identified in healthy patients or as secondary findings exist for only a few genes and conditions currently, although frameworks defining actionability have been proposed.36 Among our participants, ClinGen Actionability Working Group Evidence-based Summaries currently existed for two of the eight secondary monogenic disease risks identified in cardiology patients (both in F5), and five of the thirteen monogenic disease risks identified in primary care patients (HFE (x2), KCNQ1, TNNT2, and F5).37
Importantly, incorporating WGS into primary care and cardiology settings did not appear to increase downstream health sector costs over the short-term, even among patients who learned about a monogenic risk for disease through secondary findings. We observed more healthcare visits in both randomization arms and both cohorts in the six months following disclosure relative to the six months preceding disclosure, and in cardiology settings, we also observed more imaging and cardiology tests in both arms in the post-disclosure period. These differences are possibly a result of the study design, which mandated an in-person appointment for disclosure that physician participants often also used as well-care visits and health maintenance exams. Regardless, changes in healthcare utilization and associated costs did not differ by randomization status. The study may have been too small to detect differences, especially considering the primary analytic approach to include all healthcare services and out-of-pocket expenses. In addition, it is also possible that physicians within the MedSeq Project considered few of the WGS findings to be actionable, as evidenced by the small number of immediately attributable services we identified. If a Genome Resource Center had not been available, it is possible that we would have observed additional patient referrals to specialists, particularly among primary care patients who received WGS, and there would have been additional costs associated with these referrals. Nevertheless, the null findings around downstream health costs are notable given that the MedSeq Project’s reporting criteria that was far broader than recommendations from the American College of Medical Genetics and Genomics (ACMG). The MedSeq Project examined a far more extensive list than most research or clinical studies, examining over 4,600 genes and including findings about carrier status, pharmacogenomic information, and polygenic risk predictions. Our sensitivity analyses suggest that using more liberal reporting criteria doubled the number of potentially health-relevant findings that were reported while increasing overall costs minimally (e.g., less than $200 per patient, on average, to report secondary findings classified as likely pathogenic or uncertain significance: favor pathogenic). Numerous commentators have encouraged strict criteria for the reporting of secondary findings and discourage the use of WGS among healthy populations altogether, citing the possibility that disclosure could initiate follow-up testing and screening that increase costs while accruing limited benefits and possibly causing harms.13,38,39 Our data may begin to assuage those concerns.
Although this was a randomized controlled trial with a planned cost analysis, there are notable limitations. Physicians and patients were enrolled from a single well-supported health care system in metro Boston, Massachusetts, and physicians and patients in other settings may be less willing to undertake WGS for clinical and financial reasons. It is possible that we would have observed more use of support resources, like the Genome Resource Center, and more follow-up screening and testing if we had enrolled physicians who were less familiar with genomics or had less access to knowledgeable colleagues. As noted above, the sample size was too small to draw conclusions on a population level, and the six month time horizon may have been too short to observe the full impact on costs and health benefits. Patient time costs were not assessed, nor were the effects of disclosure on participants’ family members, precluding a complete analysis from a societal perspective. Physicians benefitted from extensive study-specific education, and were supported by a Genome Resource Center. Control arm participants received a family history review that was more rigorous than standard of care, which may have increased their downstream costs. Finally, our patient participants were more educated and less ethnically diverse than the general population. Lack of diversity has been an ongoing problem in genomic research,40 although recent efforts such as the All of Us Research Program have prioritized enrolling more representative populations. Nevertheless, our study provides novel and much-needed data to help decision makers begin to understand the short-term cost implications of integrating WGS into clinical care, and provides insight about what data are needed to provide more clarity about the economic implications of this technology.
Supplementary Material
Acknowledgments
This study was supported by NIH grants U01-HG006500, K01-HG009173, KL2-TR001100, and R01-HG007063, and Career Development Award IK2-CX001262 from the VA Clinical Sciences Research and Development Service. This work was conducted with support from Harvard Catalyst | The Harvard Clinical and Translational Science Center (National Center for Research Resources and National Center for Advancing Translational Sciences, NIH grant UL1-TR001102) and financial contributions from Harvard University and its affiliated academic health care centers. The authors thank 5AM Solutions, Inc. (Rockville, MD, USA), for their help in customizing the workflow of the “My Family Health Portrait” web tool for this study. The authors also thank Drs. Kenneth Freedberg, Marc Williams, and Ann Wu for their intellectual contributions to the design of these analyses.
Footnotes
Conflict of interest
The authors declare the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Dr. McGuire is a consultant for Human Longevity outside the submitted work. Dr. Green receives compensation for speaking or consultation from AIA, GenePeeks, Helix, Illumina, Ohana, Prudential and Veritas, and is co-founder and advisor to Genome Medical, Inc. Dr. Dukhovny reports consulting for Vermont Oxford Network, Gerson Lehrman Group, and ClearView Healthcare Partners and being faculty for Vermont Oxford Network outside the submitted work.
References
- 1.McCarthy JJ, McLeod HL, Ginsburg GS. Genomic medicine: a decade of successes, challenges, and opportunities. Sci Transl Med. 2013;5(189):189sr184–189sr184. doi: 10.1126/scitranslmed.3005785. [DOI] [PubMed] [Google Scholar]
- 2.Manolio TA. Incorporating whole-genome sequencing into primary care: falling barriers and next steps. Ann Intern Med. 2017;167(3):204–205. doi: 10.7326/M17-1518. [DOI] [PubMed] [Google Scholar]
- 3.Armstrong K. Can genomics bend the cost curve? JAMA. 2012;307(10):1031–1032. doi: 10.1001/jama.2012.261. [DOI] [PubMed] [Google Scholar]
- 4.Phillips KA, Ann Sakowski J, Trosman J, Douglas MP, Liang S-Y, Neumann P. The economic value of personalized medicine tests: what we know and what we need to know. Genet Med. 2014;16(3):251–257. doi: 10.1038/gim.2013.122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hatz MH, Schremser K, Rogowski WH. Is individualized medicine more cost-effective? A systematic review. PharmacoEconomics. 2014;32(5):443–455. doi: 10.1007/s40273-014-0143-0. [DOI] [PubMed] [Google Scholar]
- 6.Diamandis EP. The Hundred Person Wellness Project and Google's Baseline Study: medical revolution or unnecessary and potentially harmful over-testing? BMC Med. 2015;13:5. doi: 10.1186/s12916-014-0239-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Caulfield T, Evans J, McGuire A, et al. Reflections on the cost of "low-cost" whole genome sequencing: framing the health policy debate. PLoS Biol. 2013;11(11):e1001699. doi: 10.1371/journal.pbio.1001699. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Tsiplova K, Zur RM, Marshall CR, et al. A microcosting and cost-consequence analysis of clinical genomic testing strategies in autism spectrum disorder. Genet Med. 2017;19(11):1268–1275. doi: 10.1038/gim.2017.47. [DOI] [PubMed] [Google Scholar]
- 9.Sabatini LM, Mathews C, Ptak D, et al. Genomic sequencing procedure microcosting analysis and health economic cost-impact analysis: a report of the Association for Molecular Pathology. J Mol Diagn. 2016;18(3):319–328. doi: 10.1016/j.jmoldx.2015.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Sagoo GS, Norbury G, Mohammed S, Kroese M. Whole-exome sequencing in clinical genetics. A health economic evaluation. Cambridge, UK: PHG Foundation; 2017. [Google Scholar]
- 11.Stark Z, Schofield D, Alam K, et al. Prospective comparison of the cost-effectiveness of clinical whole-exome sequencing with that of usual care overwhelmingly supports early use and reimbursement. Genet Med. 2017;19(8):867–874. doi: 10.1038/gim.2016.221. [DOI] [PubMed] [Google Scholar]
- 12.Tan TY, Dillon OJ, Stark Z, et al. Diagnostic impact and cost-effectiveness of whole-exome sequencing for ambulant children with suspected monogenic conditions. JAMA pediatrics. 2017;171(9):855–862. doi: 10.1001/jamapediatrics.2017.1755. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hegde M, Bale S, Bayrak-Toydemir P, et al. Reporting incidental findings in genomic scale clinical sequencing—a clinical laboratory perspective: a report of the Association for Molecular Pathology. J Mol Diagn. 2015;17(2):107–117. doi: 10.1016/j.jmoldx.2014.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Bennette CS, Gallego CJ, Burke W, Jarvik GP, Veenstra DL. The cost-effectiveness of returning incidental findings from next-generation genomic sequencing. Genet Med. 2015;17(7):587–595. doi: 10.1038/gim.2014.156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Christensen KD, Roberts JS, Whitehouse PJ, et al. Disclosing pleiotropic effects during genetic risk assessment for Alzheimer disease. A randomized trial. Ann Intern Med. 2016;164(3):155–163. doi: 10.7326/M15-0187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Phillips KA, Pletcher MJ, Ladabaum U. Is the "$1000 genome" really $1000? Understanding the full benefits and costs of genomic sequencing. Technol Health Care. 2015;23(3):373–379. doi: 10.3233/THC-150900. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Christensen KD, Vassy JL, Jamal L, et al. Are physicians prepared for whole genome sequencing? a qualitative analysis. Clin Genet. 2016;89(2):228–234. doi: 10.1111/cge.12626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Vassy JL, Christensen KD, Schonman EF, et al. The impact of whole genome sequencing on the primary care and outcomes of healthy adult patients: a pilot randomized trial. Ann Intern Med. 2017;167(3):159–169. doi: 10.7326/M17-0188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Vassy J, Lautenbach D, McLaughlin H, et al. The MedSeq Project: a randomized trial of integrating whole genome sequencing into clinical medicine. Trials. 2014;15(1):85. doi: 10.1186/1745-6215-15-85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.McLaughlin HM, Ceyhan-Birsoy O, Christensen KD, et al. A systematic approach to the reporting of medically relevant findings from whole genome sequencing. BMC Med Genet. 2014;15:134. doi: 10.1186/s12881-014-0134-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kong SW, Lee IH, Leshchiner I, et al. Summarizing polygenic risks for complex diseases in a clinical whole-genome report. Genet Med. 2015;17(7):536–544. doi: 10.1038/gim.2014.143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Vassy JL, McLaughlin HL, MacRae CA, et al. A one-page summary report of genome sequencing for the healthy adult. Public Health Genomics. 2015;18(2):123–129. doi: 10.1159/000370102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Cirino AL, Lakdawala NK, McDonough B, et al. A comparison of whole genome sequencing to multigene panel testing in hypertrophic cardiomyopathy patients. Circulation Cardiovascular Genetics. 2017;10(5):e001768. doi: 10.1161/CIRCGENETICS.117.001768. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ramsey SD, Willke RJ, Glick H, et al. Cost-effectiveness analysis alongside clinical trials II—an ISPOR Good Research Practices Task Force report. Value Health. 2015;18(2):161–172. doi: 10.1016/j.jval.2015.02.001. [DOI] [PubMed] [Google Scholar]
- 25.Sanders GD, Neumann PJ, Basu A, et al. Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: Second panel on cost-effectiveness in health and medicine. JAMA. 2016;316(10):1093–1103. doi: 10.1001/jama.2016.12195. [DOI] [PubMed] [Google Scholar]
- 26.Frick KD. Microcosting quantity data collection methods. Med Care. 2009;47(7_Supplement_1):S76–S81. doi: 10.1097/MLR.0b013e31819bc064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.National Human Genome Research Institute. [Accessed Nov 30, 2016];The cost of sequencing a human genome. 2016 https://www.genome.gov/sequencingcosts/
- 28.Dukhovny D, Dennis CL, Hodnett E, et al. Prospective economic evaluation of a peer support intervention for prevention of postpartum depression among high-risk women in Ontario, Canada. Am J Perinatol. 2013;30(8):631–642. doi: 10.1055/s-0032-1331029. [DOI] [PubMed] [Google Scholar]
- 29.Pagon RA, Tarczy-Hornoch P, Baskin PK, et al. GeneTests-GeneClinics: Genetic testing information for a growing audience. Hum Mutat. 2002;19(5):501–509. doi: 10.1002/humu.10069. [DOI] [PubMed] [Google Scholar]
- 30.Hamosh A, Scott AF, Amberger J, Valle D, McKusick VA. Online Mendelian Inheritance in Man (OMIM) Hum Mutat. 2000;15(1):57–61. doi: 10.1002/(SICI)1098-1004(200001)15:1<57::AID-HUMU12>3.0.CO;2-G. [DOI] [PubMed] [Google Scholar]
- 31.Kalia SS, Adelman K, Bale SJ, et al. Recommendations for reporting of secondary findings in clinical exome and genome sequencing, 2016 update (ACMG SF v2.0): a policy statement of the American College of Medical Genetics and Genomics. Genet Med. 2017;19(2):249–255. doi: 10.1038/gim.2016.190. [DOI] [PubMed] [Google Scholar]
- 32.Dewey FE, Grove ME, Pan C, et al. Clinical interpretation and implications of whole-genome sequencing. JAMA. 2014;311(10):1035–1045. doi: 10.1001/jama.2014.1717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Baudhuin LM, Lagerstedt SA, Klee EW, Fadra N, Oglesbee D, Ferber MJ. Confirming variants in next-generation sequencing panel testing by Sanger sequencing. J Mol Diagn. 2015;17(4):456–461. doi: 10.1016/j.jmoldx.2015.03.004. [DOI] [PubMed] [Google Scholar]
- 34.Walsh R, Thomson KL, Ware JS, et al. Reassessment of Mendelian gene pathogenicity using 7,855 cardiomyopathy cases and 60,706 reference samples. Genet Med. 2017;19(2):192–203. doi: 10.1038/gim.2016.90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Chakravorty S, Hegde M. Gene and variant annotation for Mendelian disorders in the era of advanced sequencing technologies. Annu Rev Genomics Hum Genet. 2017;18:229–256. doi: 10.1146/annurev-genom-083115-022545. [DOI] [PubMed] [Google Scholar]
- 36.Berg JS, Foreman AK, O'Daniel JM, et al. A semiquantitative metric for evaluating clinical actionability of incidental or secondary findings from genome-scale sequencing. Genet Med. 2016;18(5):467–475. doi: 10.1038/gim.2015.104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.ClinGen Clinical Genome Resource. [Accessed Mar 17, 2017];Actionability Working Group Evidence-based Summaries. 2017 https://www.clinicalgenome.org/working-groups/actionability/projects-initiatives/actionability-evidence-based-summaries/
- 38.Yu J-H, Harrell TM, Jamal SM, Tabor HK, Bamshad MJ. Attitudes of genetics professionals toward the return of incidental results from exome and whole-genome sequencing. Am J Hum Genet. 2014;95(1):77–84. doi: 10.1016/j.ajhg.2014.06.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Kohane IS, Hsing M, Kong SW. Taxonomizing, sizing, and overcoming the incidentalome. Genet Med. 2012;14(4):399–404. doi: 10.1038/gim.2011.68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Bustamante CD, Burchard EG, De La Vega FM. Genomics for the world. Nature. 2011;475(7355):163–165. doi: 10.1038/475163a. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.