Associations are measured among all participants (n = 8270) and participants with no prior vascular disease (n = 6855) per 40-mg/dL (1-mmol/L) increment using Cox proportional hazard regression. Hazard ratios (HRs) were adjusted for age, sex, race (white vs other), randomized blood pressure treatment, randomized statin treatment, smoking, diabetes, and body mass index. HDL indicates high-density lipoprotein; LDL, low-density lipoprotein.
aCalculated using the Friedewald equation.11,12
bExcludes participants with triglyceride levels greater than 400 mg/dL (n = 370).
cCalculated using the Martin-Hopkins equation.13,14
eTable 1. Spearman Correlation Coefficients of Fasting and Nonfasting Lipids Measured on the Same Individuals
eTable 2. Associations of Fasting and Nonfasting Lipids (per SD. Increment) With Incident Coronary Events
eTable 3. Associations of Fasting and Nonfasting Lipids (per mmol/L Increment) With Incident Coronary Events in Men and Women
eTable 4. Associations of Fasting and Nonfasting Lipids (per mmol/L) With Coronary Events by Randomized Atorvastatin Therapy
eTable 5. Associations of Fasting and Nonfasting Lipids (per mmol/L) With Incident Atherosclerotic Cardiovascular Disease (ASCVD) Events (351 Events)
eTable 6. 10-fold Cross-validated Difference in the C Indices Comparing Fasting and Nonfasting Lipids in Multivariable Models Using Bootstrapping Methodology
eTable 7. Hazard Ratios (95% CIs) of Incident Coronary Events Associated With Discordant and Concordant Concentrations of Fasting and Nonfasting Lipids Based on Differences Between Nonfasting and Fasting Lipids
eTable 8. Hazard Ratios (95% CIs) of Incident Coronary Events Associated With Fasting and Nonfasting Lipids (per mmol/L) Among the Discordant Groups
eFigure 1. The ASCOT Study Design and Primary Outcome
eFigure 2. Fasting and Nonfasting Lipid Levels
Customize your JAMA Network experience by selecting one or more topics from the list below.
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.
Mora S, Chang CL, Moorthy MV, Sever PS. Association of Nonfasting vs Fasting Lipid Levels With Risk of Major Coronary Events in the Anglo-Scandinavian Cardiac Outcomes Trial–Lipid Lowering Arm. JAMA Intern Med. 2019;179(7):898–905. doi:10.1001/jamainternmed.2019.0392
How do nonfasting lipid levels compare with fasting lipid levels measured in the same individuals for assessing cardiovascular risk, and what is their association with incident cardiovascular events?
In this secondary analysis of 8270 participants in the Anglo-Scandinavian Cardiac Outcomes Trial–Lipid Lowering Arm, nonfasting lipid levels were similar to fasting lipid levels measured 4 weeks apart in the same participants in association with incident cardiovascular events overall and by randomized statin therapy. Concordance of fasting and nonfasting lipid levels for classifying participants into appropriate risk categories was high.
The present study provides robust evidence supportive of broader adoption of nonfasting lipid level measurement in clinical practice.
Recent guidelines have recommended nonfasting for routine testing of lipid levels based on comparisons of nonfasting and fasting populations. However, no previous study has examined the association of cardiovascular outcomes with fasting vs nonfasting lipid levels measured in the same individuals.
To compare the association of nonfasting and fasting lipid levels with prospectively ascertained coronary and vascular outcomes and to evaluate whether a strategy of using nonfasting instead of fasting lipid level measurement would result in misclassification of risk for individuals undergoing evaluation for initiation of statin therapy.
Design, Setting, and Participants
This post hoc prospective follow-up of a randomized clinical trial included 8270 of 10 305 participants from the Anglo-Scandinavian Cardiac Outcomes Trial–Lipid Lowering Arm (ASCOT-LLA) with nonfasting and fasting lipid levels measured 4 weeks apart (including 6855 participants with no prior vascular disease) (median follow-up, 3.3 years; interquartile range, 2.8-3.6 years). Data were collected from February 1, 1998, to December 31, 2002, and analyzed from February 1, 2016, to November 30, 2018. Multivariable Cox models, adjusted for cardiovascular risk factors, were calculated for 40-mg/dL (1-mmol/L) higher values of nonfasting and fasting lipids.
Main Outcomes and Measures
The trial’s primary end point consisted of major coronary events (nonfatal myocardial infarction [MI] and fatal coronary heart disease [212 events]). Secondary analyses examined atherosclerotic cardiovascular disease (ASCVD) events (including MI, stroke, and ASCVD death [351 events]).
Among the 8270 participants (82.1% male; mean [SD] age, 63.4 [8.5] years), nonfasting samples had modestly higher triglyceride levels and similar cholesterol levels compared to fasting samples. Associations of nonfasting lipid levels with coronary events were similar to those for fasting lipid levels. For example, adjusted hazard ratios (HRs) per 40-mg/dL of low-density lipoprotein cholesterol were 1.32 (95% CI, 1.08-1.61; P = .007) for nonfasting levels and 1.28 (95% CI, 1.07-1.55; P = .008) for fasting levels. For the primary prevention group, adjusted HRs were 1.42 (95% CI, 1.13-1.78; P = .003) for nonfasting levels and 1.37 (95% CI, 1.11-1.69; P = .003) for fasting levels. Results were consistent by randomized treatment arm (atorvastatin calcium, 10 mg/d, or placebo) and similar for ASCVD events. Concordance of fasting and nonfasting lipid levels for classifying participants into appropriate ASCVD risk categories was high (94.8%).
Conclusions and Relevance
Measurement of nonfasting and fasting lipid levels yields similar results in the same individuals for association with incident coronary and ASCVD events. These results suggest that routine measurement of nonfasting lipid levels may help facilitate ASCVD risk screening and treatment, including consideration of when to initiate statin therapy.
We live most of our lives in the nonfasting state, which is the most representative of our physiology. Nonetheless, fasting samples have been the standard for measurement of lipid profiles, which are typically measured after an 8- to 12-hour fast.1,2 Recent studies suggest that postprandial effects do not weaken, and even may strengthen, the risk associations of lipids with atherosclerotic cardiovascular disease (ASCVD).3,4 Multiple reports from well-conducted prospective studies have found similar risk associations for nonfasting and fasting lipid levels.5-9
None of these previous studies, to our knowledge, examined the association of cardiovascular events with lipid levels measured in the fasting and nonfasting states in the same individuals; instead, previous prospective studies3-9 relied on comparisons across populations of fasting and nonfasting individuals for evaluating associations with ASCVD events. Other studies compared fasting and nonfasting lipid levels cross-sectionally in the same individuals but without prospective follow-up for comparison of risk for clinical events. This difference is important because individual-level variability in lipid levels when measured in the fasting or the nonfasting state from the same individual may not be captured when looking at population-level risk associations. This unmeasured variability has been raised as a limitation of the evidence base against adopting widespread nonfasting lipid level testing.2,10 In addition to estimating cardiovascular risk and screening for dyslipidemia, another major use for lipid level testing in clinical practice is to determine eligibility for statin therapy and to tailor therapies for modifying lipid levels. Therefore, it is important to determine whether substituting nonfasting for fasting lipid level testing in the cardiovascular risk assessment currently recommended before initiating statin therapy and as part of high blood pressure management could lead to misclassification of risk.
To address this gap in the evidence, we prospectively tested individual-level differences in fasting and nonfasting lipid levels in participants from the large-scale randomized clinical Anglo-Scandinavian Cardiac Outcomes Trial–Lipid Lowering Arm (ASCOT-LLA). In ASCOT-LLA, fasting and nonfasting lipid levels were measured in the same individuals before randomization as part of the trial protocol.11,12 Herein, we compared the association of nonfasting vs fasting lipid levels with prospectively ascertained ASCVD outcomes. Furthermore, we evaluated whether a strategy of using nonfasting instead of fasting lipid levels would result in substantial misclassification of risk for individuals deemed eligible for statin therapy based on US or European guidelines.
ASCOT-LLA is a randomized, double-blinded, placebo-controlled clinical trial that tested whether atorvastatin calcium 10 mg/d or placebo in patients at high risk for ASCVD would reduce the rates of incident coronary events. The fully detailed ASCOT protocol was published previously11,12 (see eFigure 1 in the Supplement). Briefly, the study included adults aged 40 to 79 years with hypertension and total cholesterol concentrations of 250 mg/dL or lower (to convert to millimoles per liter, multiply by 0.0259) who were not taking a statin or a fibrate and who had at least 3 additional ASCVD risk factors, including left-ventricular hypertrophy, other specified abnormalities on electrocardiography, type 2 diabetes, peripheral arterial disease, previous stroke or transient ischemic attack, male sex, age of 55 years or older, microalbuminuria or proteinuria, smoking, ratio of total to high-density lipoprotein (HDL) cholesterol of 6.0 or higher, or family history of premature coronary disease.11,12 The protocol and subsequent amendments to the protocol were reviewed and ratified by central and regional ethics review boards in the UK and by national ethics and statutory bodies in Ireland and the Nordic countries. All patients provided written informed consent.
Data for this study were collected from February 1, 1998, to December 31, 2002. For this post hoc secondary analysis, we included a total of 8270 of the 10 305 participants randomized into ASCOT-LLA who had nonfasting and fasting lipid levels available for analysis. These measurements were taken using the trial protocol approximately 4 weeks apart before randomization, with no intervention or advice given between the 2 visits.11,12 At the screening visit that occurred approximately 4 weeks before randomization, relevant characteristics of the patients were assessed, their eligibility for the trial was determined, written informed consent was obtained, and nonfasting blood samples were collected. The randomization visit approximately 4 weeks later included the collection of fasting blood samples, with fasting defined as no food or drink except plain water for at least 8 hours before the blood draw. Two central laboratories, one located in Ireland (for the United Kingdom and Ireland) and the other located in Sweden (for the Nordic countries), analyzed blood samples for levels of total cholesterol, HDL cholesterol, and triglycerides. Low-density lipoprotein (LDL) cholesterol was calculated using the Friedewald equation11,12 and the novel Martin-Hopkins equation.13,14
The trial primary end point of major coronary events was a composite consisting of nonfatal myocardial infarction (MI) and fatal coronary heart disease. In secondary analyses, we also examined the expanded composite end point of ASCVD events defined as fatal or nonfatal MI, fatal or nonfatal stroke, or ASCVD death.
Data were analyzed from February 1, 2016, to November 30, 2018. We compared participants with and without coronary events during follow-up using unpaired two-tailed t tests or Kruskal-Wallis tests for continuous variables and χ2 tests for categorical baseline lipid and nonlipid variables. We calculated Spearman correlation coefficients and 95% CIs between fasting and nonfasting lipid levels.
We performed unadjusted and multivariable Cox proportional hazards regression with models that adjusted for cardiovascular risk factors, including age, sex, race (white vs other), smoking status, diabetes, randomized blood pressure treatment (atenolol or amlodipine besylate), randomized statin treatment or placebo, and body mass index.
We investigated associations between each lipid variable (fasting or nonfasting) per 40-mg/dL (1-mmol/L) higher value with incident major coronary and ASCVD events using separate models. We repeated selected analyses per SD increment for comparison with prior studies that reported results per SD increment. Because ASCOT-LLA included a small proportion of participants with prior vascular disease, we repeated the analyses after excluding these participants to examine risk associations in an exclusively primary prevention population. We also performed sensitivity analyses after excluding participants with triglyceride levels greater than 400 mg/dL (n = 370) (to convert to millimoles per liter, multiply by 0.0113). We tested for statistical interaction between randomized statin assignment or sex and each of the lipid variables in association with outcomes using likelihood ratio tests. Subsequently, we used 10-fold cross-validated differences in Harrell C indices15 comparing fasting and nonfasting lipid levels in multivariable risk factor–adjusted models using bootstrapping methods to evaluate for statistically significant differences between fasting and nonfasting models.
To examine the extent of agreement between the associations of fasting and nonfasting lipid levels with outcomes, we calculated ASCVD risks using the 2013 American College of Cardiology/American Heart Association (ACC/AHA) pooled cohorts1 and the QRISK2 equations (QRESEARCH, University of Nottingham [https://www.qresearch.org/]). Participants were classified into low-, moderate-, and high-risk categories using the recommended clinical categories of less than 5.0%, 5.0% to less than 7.5%, and at least 7.5% for the ACC/AHA risk scores and less than 10.0%, 10.0% to less than 20.0%, and at least 20.0% for the QRISK2 risk score. The degree of misclassification of individuals into coronary and ASCVD risk categories, respectively, was assessed. Finally, we also examined associations with incident coronary events for participants with discordant or concordant concentrations based on within-individual differences between their nonfasting and fasting lipid levels. Two-tailed P < .05 was considered significant. Statistical analyses were performed with SAS, version 9.4 (SAS Institute, Inc).
The ASCOT-LLA trial was terminated prematurely after a median follow-up of 3.3 years (interquartile range, 2.8-3.6 years) owing to a highly significant relative risk reduction (HR, 0.64; 95% CI, 0.50-0.83; P < .001) in favor of atorvastatin calcium 10 mg/d compared with placebo on the primary coronary end point (eFigure 1 in the Supplement).12 During the trial, a total of 212 incident major coronary events (the trial primary end point) occurred among the 8270 participants who had fasting and nonfasting lipids measured at baseline and were included in the analysis (6791 men [82.1%] and 1479 women [17.9%]; mean [SD] age, 63.4 [8.5] years). Compared with those participants without coronary events, those who had a coronary event were significantly older (mean [SD] age, 65.6 [8.7] vs 63.3 [8.5] years; P < .001), were more likely to have a history of stroke or transient ischemic attack (30 of 212 [14.2%] vs 801 of 8058 [9.9%]; P = .04) or diabetes (77 of 212 [36.3%] vs 2123 of 8058 [26.3%]; P = .001), and had more cardiovascular risk factors (median, 4 [interquartile range, 3-5] vs 3 [interquartile range, 3-4]; P < .001) (Table 1). Compared with fasting samples, nonfasting samples obtained from the same individuals had modestly higher triglyceride levels and negligible differences in total, LDL, and HDL cholesterol levels (Table 1 and eFigure 2 in the Supplement). Spearman correlation coefficients between fasting and nonfasting lipids ranged from 0.693 for triglyceride level to 0.878 for HDL cholesterol level (P < .001 for all) (eTable 1 in the Supplement).
Nonfasting and fasting levels of LDL and non-HDL cholesterol and the ratio of total to HDL cholesterol were positively associated and HDL cholesterol level was inversely associated with coronary events in unadjusted and risk factor–adjusted models (Figure and Table 2). Notably, the nonfasting lipid associations were similar in magnitude to fasting lipid associations, including for participants with no prior vascular disease (ie, the exclusively primary prevention group), with no significant interaction by fasting status noted for any lipid variable. For example, adjusted HRs per 40 mg/dL for Friedewald LDL cholesterol were 1.32 (95% CI, 1.08-1.61; P = .007) for nonfasting and 1.28 (95% CI, 1.07-1.55; P = .008) for fasting levels. For the exclusively primary prevention group, adjusted HRs were 1.42 (95% CI, 1.13-1.78; P = .003) for nonfasting levels and 1.37 (95% CI, 1.11-1.69; P = .003) for fasting levels (P = .85 and P = .003, respectively, for interaction by fasting status). Sensitivity analysis showed a similar significant result in Friedewald LDL cholesterol level after excluding triglyceride levels of greater than 400 mg/dL. We obtained similar results when LDL cholesterol level was calculated using the novel Martin-Hopkins equation with some strengthening of the magnitude of association for nonfasting LDL cholesterol (HR for nonfasting level, 1.39 [95% CI, 1.11-1.73; P = .004]; HR for fasting level, 1.31 [95% CI, 1.08-1.58; P = .007]). For the exclusively primary prevention group, adjusted HR for nonfasting level was 1.53 (95% CI, 1.19-1.97; P = .001), and HR for fasting level was 1.41 (95% CI, 1.13-1.75; P = .002) (P = .69 and P = .62, respectively, for interaction by fasting status). Similar results were obtained for analyses by SD increments (eTable 2 in the Supplement). We found no evidence of interaction between sex and each of the study lipid levels in association with the risk of coronary events (eTable 3 in the Supplement).
The associations of nonfasting and fasting lipid levels with coronary events were similar according to randomized atorvastatin calcium therapy (10 mg/d) vs placebo and consistent with the overall study (eTable 4 in the Supplement). No statistically significant interactions were noted between randomized atorvastatin treatment and fasting status in association with risk of coronary events. For consistency with clinical prevention guidelines, we also examined fasting and nonfasting lipid associations with incident ASCVD events (MI, stroke, or ASCVD death [n = 315]) and found similar results to the primary coronary end point.
The overall concordance of fasting and nonfasting lipid levels for classifying participants into categories of ASCVD risk was very high (Table 3), whether risk was calculated based on the 2013 ACC/AHA pooled cohort risk equations (94.8% concordance) or the QRISK2 algorithm (98.6% concordance). We also found highly concordant results among the participants with fasting or nonfasting LDL cholesterol levels of 70 to 189 mg/dL, which is the LDL cholesterol range of the ACC/AHA cholesterol guidelines that are used for statin eligibility. Moreover, no differences occurred in the proportion of reclassification from high-risk to lower-risk categories between fasting and nonfasting lipids for the ACC/AHA guidelines (1.41% vs 1.40%) or QRISK2 algorithms (0.42% vs 0.45%).
eTable 6 in the Supplement shows no statistically significant differences comparing the 10-fold cross-validated difference in the C indices between fasting and nonfasting lipid levels in multivariable models using bootstrapping methods. Finally, to address the possibility that some individuals may have substantial differences in fasting and nonfasting lipid levels, which may affect treatment decisions, we classified participants based on within-individual differences between their nonfasting and fasting lipid levels as discordant in the top quartile of the distribution of differences (fasting greater than nonfasting lipid levels), concordant (25th to 75th percentile), or discordant in the bottom quartile of the distribution (fasting less than nonfasting lipid levels). For all lipid measures, the HRs were not significantly different between the 2 discordant groups (fasting greater than nonfasting and fasting less than nonfasting lipid levels) (eTable 7 in the Supplement). Moreover, no significant differences in risk associated with coronary events occurred between fasting and nonfasting lipid levels within each discordant group (eTable 8 in the Supplement). Similar results were also noted between the 2 discordant groups, except for the ratio of total to HDL cholesterol (adjusted HRs for fasting less than nonfasting levels, 5.25 [95% CI, 1.58-17.40; P = .02] for fasting and 4.00 [95% CI, 1.14-14.00; P = .06]; P = .76 for nonfasting).
In ASCOT-LLA, nonfasting lipid levels were similar to fasting lipid levels measured in the same individuals for association with incident coronary and vascular events. Results were similar by randomized allocation to atorvastatin calcium 10 mg/d or placebo. Furthermore, the very high agreement (94.8%) of ASCVD risk classification categories for fasting and nonfasting samples suggests that risk assessment and treatment decisions for statin and antihypertensive therapies would be consistent whether lipid levels were measured in the fasting or nonfasting state. Therefore, the present ASCOT-LLA results provide robust evidence and more impetus for physicians to more broadly adopt nonfasting measurement of lipid levels for routine practice, in a manner consistent with the recent guideline recommendations that have endorsed measuring nonfasting lipid levels. Such a strategy would be highly effective and offer many advantages for cardiovascular risk screening and treatment, including for initiating statin therapy and for blood pressure assessment and management decisions that rely on calculating ASCVD risk.
Since the 1970s, numerous well-conducted, large, representative case-control and prospective studies with medium- to long-term follow-up5-9 have found that the associations of lipid levels with cardiovascular events were generally consistent among groups of individuals who had fasting lipid levels measured compared with groups of individuals (albeit not the same individuals) who had nonfasting lipid levels measured. These studies have examined clinical outcomes ranging from incident ASCVD events (MI, stroke, and revascularization) to cardiovascular or all-cause mortality, finding consistent associations for nonfasting lipid profiles with risk of cardiovascular events. Moreover, some studies that included fasting and nonfasting populations found stronger risk associations for nonfasting lipid levels such as triglycerides.6,7 This evidence base prompted the 2016 and 2018 European Atherosclerosis Society and the European Federation of Laboratory Medicine Consensus Statements3,16,17 to recommend using nonfasting lipid levels for routine cardiovascular care, including for cardiovascular risk assessment and treatment decisions. Nonfasting lipid levels have also been recently endorsed by Canadian guidelines,18,19 several other international guidelines,20,21 and more recently by the 2018 ACC/AHA cholesterol guidelines.22
The present study is, to our knowledge, the first to compare prospective associations of fasting and nonfasting lipid levels measured in the same individuals with incident vascular events. Consistent with prior studies that were performed in populations of fasting and nonfasting individuals,5-7,9,23 we found small, clinically nonmeaningful differences between fasting and nonfasting lipid levels measured in the same individuals, with modestly greater nonfasting triglyceride levels, slightly lower nonfasting LDL cholesterol levels, and no differences in HDL cholesterol levels. One major limitation to more widespread acceptance of nonfasting lipid levels has been that prior studies that evaluated ASCVD risk associations for fasting and nonfasting lipids measured them in groups of fasting and nonfasting individuals, but not in the same individuals.2,10 This potential concern is relevant because physicians treat individual patients (not populations), and individual-level variability in fasting vs nonfasting lipid levels may not be captured when looking at population-level risk associations. In this regard, the present study provides new and robust evidence that, even when measured in the same individuals, nonfasting lipid levels are at least as good as fasting lipid levels for calculating cardiovascular risk, including for guiding risk classification and preventive treatment decisions.
Furthermore, recent studies suggest that fasting for lipid level testing may be risky in elderly patients or those with diabetes who use hypoglycemic medications. Approximately 1 of 4 patients with diabetes reported having an en-route hypoglycemic event owing to fasting for blood tests.24,25 These events are underreported and add to patient morbidity. Hence, a strategy that measures lipid levels without requiring fasting after normal food intake has numerous practical advantages for clinical practice.4,10,20
Strengths of the present study include that nonfasting and fasting lipid levels were measured in the same individuals 4 weeks apart according to a centralized trial protocol with no intervention or advice during the 4-week period. Other strengths included the large ASCOT-LLA study population with individuals undergoing primary and secondary prevention and detailed information collected on cardiovascular risk factors, the prospectively adjudicated clinical outcomes in this clinical trial, and the randomized allocation of statin therapy vs placebo. Potential limitations include that the ASCOT-LLA trial inclusion criteria may limit generalizability of the findings, although our results are consistent with prior population-based studies that compared group differences in fasting and nonfasting lipid levels. ASCOT-LLA included patients with and without prior vascular disease, which may make the results more generalizable. Because the Friedewald equation has recently been recognized to have limitations,13 especially in individuals with higher triglyceride levels or lower LDL cholesterol levels, we also repeated the analyses for LDL cholesterol level calculated by the novel Martin-Hopkins equation,13,14 finding similar results. Finally, few nonwhite individuals were enrolled in ASCOT-LLA, and future research should assess potential ethnic and/or racial differences.
In ASCOT-LLA, association with cardiovascular risk was at least as good for nonfasting as for fasting lipid levels measured in the same individuals, and lipid level measurements had very high concordance for cardiovascular risk categorization regardless of fasting status. These results are consistent with prior population-level studies that examined fasting and nonfasting lipid levels and support the recent change in US guidelines22 to more broadly adopt nonfasting lipids for routine cardiovascular risk assessment. The results of this study suggest that such a strategy would be highly effective and offer many advantages for cardiovascular risk screening and treatment decisions, including for initiating statin or antihypertensive therapy.
Accepted for Publication: February 3, 2019.
Corresponding Author: Samia Mora, MD, MHS, Center for Lipid Metabolomics, Division of Preventive Medicine, Brigham and Women’s Hospital, Harvard Medical School, 900 Commonwealth Ave E, Boston, MA 02215 (email@example.com).
Published Online: May 28, 2019. doi:10.1001/jamainternmed.2019.0392
Author Contributions: Dr Sever had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Mora, Sever.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Mora, Chang, Sever.
Critical revision of the manuscript for important intellectual content: Mora, Moorthy.
Statistical analysis: Chang, Moorthy.
Obtained funding: Sever.
Administrative, technical, or material support: Sever.
Supervision: Mora, Sever.
Conflict of Interest Disclosures: Dr Mora reported receiving grants from Pfizer, the National Heart, Lung, and Blood Institute, and the National Institute of Diabetes and Digestive and Kidney Diseases during the conduct of the study; personal fees from Pfizer, Amgen, and Quest Diagnostics, and grants from Atherotech Diagnostics Lab outside the submitted work. Dr Sever reported receiving research grant support and personal fees from Pfizer and Amgen outside the submitted work. No other disclosures were reported.
Funding/Support: This study was supported by an investigator-initiated institutional research grant from Pfizer; grants K24HL136852, R01HL134811, and HL117861 from the National Heart, Lung, and Blood Institute of the National Institutes of Health (Dr Mora); a Senior Investigator Award from the National Institute for Health Research (Dr Sever); and a Biomedical Research Centre Award to Imperial College from the Healthcare National Health Service Trust (Dr Sever). The ASCOT trial was financially supported by Pfizer.
Role of the Funder/Sponsor: The sponsors 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: The content and opinions expressed in the manuscript are solely the responsibility of the study authors and do not necessarily represent the views of the US National Institutes of Health or the UK National Health Service.