Association between pre-operative statin use and major cardiovascular complications among patients undergoing non-cardiac surgery: the VISION study

Abstract

Aims

The aim of this study was to assess the effects of pre-operative statin therapy on cardiovascular events in the first 30-days after non-cardiac surgery.

Methods and results

We conducted an international, prospective, cohort study of patients who were ≥45 years having in-patient non-cardiac surgery. We estimated the probability of receiving statins pre-operatively using a multivariable logistic model and conducted a propensity score analysis to correct for confounding. A total of 15 478 patients were recruited at 12 centres in eight countries from August 2007 to January 2011. The matched population consisted of 2845 patients (18.4%) treated with a statin and 4492 (29.0%) controls. The pre-operative use of statins was associated with lower risk of the primary outcome, a composite of all-cause mortality, myocardial injury after non-cardiac surgery (MINS), or stroke at 30 days [relative risk (RR), 0.83; 95% confidence interval (CI), 0.73–0.95; P = 0.007]. Statins were also associated with a significant lower risk of all-cause mortality (RR, 0.58; 95% CI, 0.40–0.83; P = 0.003), cardiovascular mortality (RR, 0.42; 95% CI, 0.23–0.76; P = 0.004), and MINS (RR, 0.86; 95% CI, 0.73–0.98; P = 0.02). There were no statistically significant differences in the risk of myocardial infarction or stroke.

Conclusion

Among patients undergoing non-cardiac surgery, pre-operative statin therapy was independently associated with a lower risk of cardiovascular outcomes at 30 days. These results require confirmation in a large randomized trial.

Clinical trial registration

Clinical Trials.gov NCT00512109

See page 186 for the editorial comment on this article (doi:10.1093/eurheartj/ehv533)

Introduction

Among the 200 million adults worldwide who undergo non-cardiac surgery annually, ≥10 million will suffer a cardiovascular complication in the first 30 days after surgery.^1,2 Despite the magnitude of the problem, no intervention has been shown to be both safe and efficacious in the prevention of cardiovascular events.^3–5

Observational studies and small randomized controlled trials (RCTs) suggest that statins may reduce the risk of cardiovascular events in patients undergoing non-cardiac surgery.^6–10 There remains, however, uncertainty regarding the effects of statins in this setting.

We recently completed recruitment to the Vascular events In non-cardiac Surgery patIents cOhort evaluatioN (VISION) study, a large international prospective cohort study evaluating perioperative events. One of its objectives was to identify promising interventions that might reduce the incidence of perioperative complications for testing in subsequent RCTs. Moreover, the VISION study allows for determination of risk-adjusted estimates of the effectiveness of preventive interventions in a representative sample of real-world patients undergoing non-cardiac surgery. Thus, for the current analyses, our objective was to assess the effects of perioperative statin usage on cardiovascular events at 30 days.

Methods

Study design

The VISION study (NCT00512109) methods were published previously.^11,12 In brief, VISION was a prospective cohort study of patients undergoing non-cardiac surgery. VISION completed the recruitment of over 40 000 patients in North and South America, Africa, Asia, Australia, and Europe. The Research Ethics Board at each site approved the protocol prior to patient recruitment. The first 16 081 patients enrolled had fourth-generation troponin T (TnT) measurements after surgery and experienced event rates three times higher than expected, which allows sufficient statistical power to address some of the VISION pre-specified questions.¹¹ This publication addresses the effects of pre-operative statin use on clinical outcomes at 30 days after surgery and is restricted to the period of fourth-generation assay use for measurement of TnT concentration. We expect to present data later this year for the remaining 25 000 patients who underwent measurement of post-operative TnT using a fifth-generation high-sensitivity assay.

Patients

Eligible patients had non-cardiac surgery requiring general or regional anaesthesia and were ≥45 years of age. We excluded patients not requiring an overnight stay after surgery, previously enrolled, or who refused consent.

Patients were identified by screening daily lists in pre-operative clinics, daily surgical lists and lists from the previous day, lists on surgical wards and in intensive care units, and patients in the pre-operative holding area. In some centres, surgical volume exceeded the capacity to enrol all eligible consecutive patients. In these centres, we created a recruitment schedule consisting of random weeks for recruitment or randomly selected surgical services.

Patients consented to participate prior to surgery or, for patients from whom we could not obtain consent pre-operatively (e.g. emergency night-time surgical case), consent was obtained within the first 24 h after surgery. Eight centres used a deferred process for patients unable to consent (e.g. patients sedated and mechanically ventilated) and no substitute decision-maker was available.

Procedures

Throughout patients’ hospital stay, researchers evaluated patients, reviewed charts, and recorded outcomes. Data regarding the use of cardiovascular drugs was collected before (>24 h to 7 days, ≤24 h) and after surgery (first 3 days). At each site, an investigator reviewed all data. Patients had blood collected to measure a Roche fourth-generation Elecsys™ TnT assay 6–12 h post-operatively and on the first 3 days after surgery. Centres submitted case report forms and supporting documentation directly to the data management system (iDataFax™, coordinating centre, McMaster University, Canada).

Data quality

Monitoring in VISION consisted of central data consistency checks, statistical monitoring, and on-site monitoring for all centres. For the on-site monitoring, the central co-ordinator randomly selected participants with and without outcomes, and monitors audited their hospital charts and supporting documents.¹¹

Outcomes

The primary outcome was a composite of all-cause mortality, myocardial injury after non-cardiac surgery (MINS), or stroke at 30 days after surgery.¹² Secondary outcomes included individual components of the primary endpoint, myocardial infarction, cardiovascular and non-cardiovascular death, sepsis, and pneumonia. Diagnostic criterion for MINS was any peak TnT ≥ 0.03 ng/mL judged to result from myocardial ischaemia (i.e. without evidence of a non-ischaemic aetiology for the TnT elevation).¹² Diagnosis of myocardial infarction was based upon the Universal Definition of Myocardial Infarction.¹³ Stroke was defined as a new focal neurological deficit thought to be vascular in origin with signs and symptoms lasting more than 24 h. All events were adjudicated.

Statistical analysis

A statistical analysis plan was written before undertaking the current analyses. Continuous and categorical variables were compared using parametric or non-parametric methods as appropriate.

To estimate the effect of pre-operative statin usage on the outcomes, we undertook propensity score analyses. We estimated the probability of patients receiving a statin pre-operatively (i.e. any usage during the 7 days before surgery) using a multivariable logistic regression model in which 27 pre-operative variables were included (Table 1). Model discrimination was assessed using the c-statistic, and calibration using a graphical representation and an LOESS algorithm.¹⁴

Table 1.

Pre-operative patient characteristics and type of surgery

	Unmatched population					Matched population
	Statin N = 3760		No Statin N = 11 718		ASD (%)	Statin N = 2845		No Statin N = 4492		ASD (%)
	n	%	n	%		n	%	n	%
Demographics
Male sex	2013	53.5	5465	46.6	13.8	1392	48.9	2077	46.2	5.4
Age (years) ± SD	69.5 ± 10.2		64.6 ± 12.1		43.9	68.8 ± 10.2		68.6 ± 11.7		1.9
Medical history
Current atrial fibrillation	190	5.0	339	2.9	12.9	141	5.0	211	4.7	1.2
History of
Congestive heart failure	315	8.4	420	3.6	25.8	197	6.9	260	5.8	4.9
Coronary artery disease	1186	31.5	699	6.0	108.0	560	19.7	609	13.6	17.9
Recent high-risk CAD	99	2.6	85	0.7	22.5	47	1.6	62	1.4	2.3
Stroke	498	13.2	637	5.4	34.4	295	10.4	401	8.9	5.1
Sleep apnoea	328	8.7	459	3.9	24.8	210	7.4	315	7.0	1.4
Peripheral vascular disease	444	11.8	389	3.3	47.4	220	7.7	233	5.2	11.5
Hypertension	2903	77.2	4986	42.5	70.1	2061	72.4	3095	68.9	7.6
COPD	430	11.4	877	7.5	15.0	307	10.8	441	9.8	3.3
Diabetes	1337	35.6	1680	14.3	60.6	852	30.0	1129	25.1	11.1
Active cancer	450	12.0	1529	13.0	3.2	369	13.0	587	13.1	0.3
Dialysis	73	1.9	97	0.8	12.3	47	1.6	61	1.4	2.5
Creatinaemia > 170 µmol L−1	215	5.7	318	2.7	18.5	137	4.8	187	4.2	3.3
Pre-operative treatment
Aspirin	1312	34.9	987	8.4	95.3	701	24.6	839	18.7	15.3
ACEI/ARB	2264	60.2	2890	24.7	82.5	1516	53.3	2162	48.1	10.3
Beta-blocker	1236	32.9	1496	12.8	60.2	721	25.3	981	21.8	8.5
Alpha-2 agonist	18	0.5	85	0.7	2.9	12	0.4	22	0.5	1.0
Rate controlling CCB	232	6.2	283	2.4	24.5	148	5.2	184	4.1	5.6
Dihydropyridine CCB	850	22.6	1370	11.7	34.0	594	20.9	814	18.1	7.2
Surgery
Urgent surgery	89	2.4	370	3.2	4.5	63	2.2	100	2.2	0.1
Emergent surgery	273	7.3	1499	12.8	16.6	225	7.9	357	8.0	0.1
Thoracic aorta reconstruction	15	0.4	12	0.1	9.3	11	0.4	6	0.1	6.9
Aorto-iliac reconstruction	90	2.4	45	0.4	32.5	32	1.1	40	0.9	2.5
Peripheral vascular reconstruction	122	3.2	74	0.6	33.0	58	2.0	56	1.2	7.1
Extracranial cerebrovascular	71	1.9	18	0.1	44.3	23	0.8	17	0.4	7.0
Endovascular aneurysm repair	46	1.2	22	0.2	23.9	15	0.5	16	0.4	2.8
Complex visceral resection	68	1.8	319	2.7	5.6	60	2.1	108	2.4	1.9
Colectomy or stomach surgery	237	6.3	824	7.0	2.8	176	6.2	279	6.2	0.1
Other intra-abdominal	234	6.2	1312	11.2	15.8	194	6.8	308	6.9	0.1
Major head/neck resection	51	1.4	216	1.8	3.6	46	1.6	74	1.6	0.2
Pneumonectomy	3	0.1	12	0.1	0.7	3	0.1	7	0.2	1.3
Lobectomy	42	1.1	127	1.1	0.3	33	1.2	57	1.3	1.0
Other thoracic	43	1.1	171	1.5	2.6	34	1.2	62	1.4	1.6
Visceral resection	142	3.8	292	2.5	8.2	100	3.5	168	3.7	1.2
Cytoreductive surgery	28	0.7	118	1.0	2.6	24	0.8	42	0.9	0.9
Hysterectomy	103	2.7	537	4.6	8.8	87	3.1	136	3.0	0.2
Radical hysterectomy	22	0.6	116	1.0	4.1	19	0.7	24	0.5	1.8
Radical prostatectomy	88	2.3	193	1.6	5.4	66	2.3	108	2.4	0.5
Transurethral prostatectomy	126	3.3	279	2.4	6.4	91	3.2	137	3.0	0.9
Major hip/pelvic	359	9.5	993	8.5	3.9	291	10.2	486	10.8	1.9
Femur osteosynthesis	70	1.9	336	2.9	6.0	61	2.1	108	2.4	1.7
Knee arthroplasty	460	12.2	859	7.3	18.8	351	12.3	566	12.6	0.8
Amputation
Above-knee	16	0.4	55	0.5	0.6	12	0.4	25	0.6	1.8
Lower-leg	35	0.9	39	0.3	10.4	21	0.7	25	0.6	2.4
Craniotomy	89	2.4	355	3.0	3.9	68	2.4	113	2.5	0.8
Major spine	119	3.2	346	2.9	1.2	96	3.4	141	3.1	1.3
Low-risk surgeries	1294	34.4	4802	41.0	13.3	1034	36.3	1625	36.2	0.3

ASD, absolute standardized difference; recent high-risk CAD, diagnosis ≤6 months prior to surgery of myocardial infarction, acute coronary syndrome, Canadian Cardiovascular Society Class (CCSC) III angina or CCSC IV angina; COPD, chronic obstructive pulmonary disease; ACEI/ARB, angiotensin-converting enzyme inhibitors/angiotensin-receptor blockers; CCB, calcium-channel blockers.

We then conducted a nearest neighbour propensity score matching with a caliper of 20% of the standard deviation of the logit of the probability of taking a statin before surgery (i.e. the propensity score).¹⁵ Patients who used a statin (‘Statin’ group) before surgery were matched with up to two patients who did not receive a statin (‘No Statin’ group). No replacement was performed, and patients from the ‘No Statin’ group without matching patients from the ‘Statin’ group were excluded.

After matching, the balance of covariates between the groups was assessed using the standardized differences expressed as a percentage for each covariate,¹⁶ and globally using c-statistic (where a value of 0.5 indicates perfect balance).¹⁷ Baseline covariates with an absolute standardized difference >10% were considered to be imbalanced.¹⁶ Relative risks (RR) and the 95% confidence intervals (CI) for all outcomes were computed using conditional Poisson regression in which imbalanced variables were included.

For the primary outcome, we undertook post hoc subgroup analyses based upon the following pre-operative conditions: coronary artery disease, stroke, peripheral vascular disease, diabetes, and renal insufficiency. We developed Cox proportional hazards models in which the dependent variable was all-cause mortality at 30 days. Unadjusted hazard ratios (HRs) were estimated in the unmatched population. Imbalanced variables were added to the model in the matched population to estimate the adjusted HRs.

We tested the impact of this stochastic procedure on the treatment effect estimation in a sensitivity analysis using 500 replications of the main analysis. We also conducted a second sensitivity analysis for the primary outcome in which statin users were defined as patients receiving statins before surgery and within the first 3 days after surgery.

P-values were two-tailed, and values <0.05 were considered significant. Statistical analyses were carried out using R software (version 3.2, http://www.r-project.org), and the MatchIt package (version 2.4-21, http://gking.harvard.edu/matchit).

Results

Patients

Figure 1 shows the patient flowchart. From the 16 081 enrolled patients, 603 were excluded from the statin analysis, mostly due to failure to measure at least one troponin concentration. Of the 15 478 included patients, 99.7% completed the follow-up. Patients were recruited at 12 centres in eight countries, in North and South America, Australia, Asia, and Europe, from August 2007 to January 2011. A total of 3760 (24.3%) patients were treated pre-operatively with a statin. Table 1 presents the baseline characteristics. Several patients not treated with a statin would be classified as high cardiovascular risk according to current guidelines¹⁸ (i.e. 13.6, 8.9, 5.2, and 25.1% had coronary disease, stroke, peripheral vascular disease, and diabetes, respectively). Vascular risk factors were more frequent in the statin group.

Figure 1.

Patient flowchart.

Matching

The model to predict treatment with a pre-operative statin demonstrated a c-statistic of 0.84. The graphical representation (Supplementary material online, Figure S1) shows excellent calibration throughout the full range of the observed probabilities for receiving a statin.

We included 2845 patients (75.6%) who were treated with a statin in the matched population; the remaining 915 patients (24.4%) were discarded, because there was no matching control. Among the 11 718 patients who were not treated with a statin, 4492 (38.3%) were included as matched controls.

Despite a large reduction in the imbalances observed in the unmatched compared with the matched populations (Table 1), absolute standardized differences remained >10% in the matched populations for: coronary artery disease, pre-operative aspirin, peripheral vascular disease, diabetes, and pre-operative angiotensin-converting enzyme inhibitor, or angiotensin-receptor blocker. In the matched population, these characteristics were more common in the Statin group. Post-matching c-statistic was 0.58, suggesting a small residual imbalance. Variables with standardized differences >10% were then included in the models used to estimate the effect of statins.

Outcomes

The 30 day primary outcome (all-cause mortality, MINS, or stroke) was observed in 1614 patients in the matched cohort (11.8%), Supplementary material online, Table S1. After adjustment for potential confounding factors, statin use was associated with a significant lower risk of the primary outcome (RR, 0.83; 95% CI, 0.73–0.95; P = 0.007), Figure 2. This relative effect corresponded to an absolute risk reduction of 2.0% (95% CI, 0.5–3.2%; P = 0.005).

Figure 2.

Effects of statins on primary and secondary outcomes. RR, relative risk; CI, confidence interval; MINS, Myocardial Injury after Non-cardiac Surgery.

The effects of statins on secondary outcomes are presented in Figure 2. Statin therapy was significantly associated with lower risk of all-cause mortality (RR, 0.58; 95% CI, 0.40–0.83; P = 0.003), cardiovascular mortality (RR, 0.42; 95% CI, 0.23–0.76; P = 0.004), and MINS (RR, 0.86; 95% CI, 0.73–0.98; P = 0.02). There were no statistically significant differences in the risk of non-cardiovascular mortality, myocardial infarction, or stroke with statin therapy. Supplementary material online, Table S1 shows the incidence of events in both groups. Statin therapy was associated with lower mortality (HR, 0.57; 95% CI, 0.47–0.69; P = 0.004) in time-dependent analyses (Figure 3). Statins were also associated with a reduction in sepsis (RR, 0.81; 95% CI, 0.71–0.92; P = 0.002) and pneumonia (RR, 0.71; 95% CI, 0.52–0.98, P = 0.038).

Figure 3.

Effect of statins on survival at 30 days. Cox proportional hazards model for the (A) unmatched and (B) matched populations. HR, hazard ratio risk; CI, confidence interval.

Subgroup analysis

Figure 4 shows the results of the subgroup analyses. There was evidence of a subgroup effects only when we compared diabetic vs. non-diabetic patients.

Figure 4.

Subgroup analysis. RR, relative risk; CI, confidence interval.

Sensitivity analysis

Among 500 replications of the main analysis, only two of them (0.4%) had a P-value greater than 0.05. Further, the median of the distribution of the RR observed among the 500 replications (Figure 5) was similar to the results of our main analysis (respectively, 0.84 and 0.83), suggesting that the main analysis was not related to a specific matching procedure.

Figure 5.

Sensitivity analysis replicating the main analysis for the primary outcome. RR, relative risk.

The second sensitivity analysis included 1653 patients who received a statin before surgery and during the first 3 days after surgery and 2847 matched controls. This analysis demonstrated a lower risk of the primary outcome (RR, 0.82; 95% CI, 0.68–0.98; P = 0.029) with statins. After 500 replications of this analysis, 83 (16.6%) were associated with a P ≥ 0.05.

Discussion

In this international cohort study of >15 000 patients undergoing non-cardiac surgery, we found that pre-operative statin use was associated with lower risk of the primary outcome at 30 days. This beneficial effect was driven mainly by statistically significant lower risks of death and MINS. We also found that statins were associated with a lower risk of cardiovascular mortality, sepsis, and pneumonia. Another message from our results is that the use of long-term statins is sub-optimal in high cardiovascular risk patients, who should be on long-term lipid-lowering therapy independently of surgery.¹⁸

The VISION results are consistent with other observational studies that evaluated statins in the perioperative context.^19–22 A systematic review that included 10 observational studies of patients undergoing non-cardiac surgery found a 30% reduction in death or myocardial infarction and a 31% reduction in mortality with statins.⁶ Despite these findings, it should be acknowledged that the majority of the studies were retrospective, single centre, and restricted to vascular surgery.

A few RCTs have evaluated the effects of statins in non-cardiac surgery. A Cochrane review pooled the results from three vascular surgery trials (total participants, 178) and found a non-significant decrease in the risk of mortality and myocardial infarction at 30 days with statins.²³ The authors considered that the evidence was insufficient to draw conclusions of whether statins resulted in either lower or higher perioperative risks. Another systematic review found a decrease in mortality and myocardial infarction with statins in patients undergoing non-cardiac surgery, but, due to the few events observed, concluded that there were insufficient data to support recommendations.²⁴ A more recent review reached similar conclusions.²⁵

Although trials in non-interventional settings have demonstrated that the benefits of statins typically do not become apparent until 6 months after initiation, trials in acute interventional settings suggest that statins may have benefits within 30 days. A meta-analysis that included 13 RCTs and 3341 percutaneous coronary intervention patients demonstrated a 44% risk reduction in cardiac events at 30 days with high-dose statins.²⁶

We observed that statins were associated with a reduction in overall and cardiovascular mortality, and MINS. Although the association for myocardial infarction was not significant, it had a similar point estimate of effect to MINS. There were substantially fewer myocardial infarctions (3.6%) compared with MINS (10.6%). This may explain why myocardial infarction had a non-significant association. We have previously demonstrated that MINS is a strong independent predictor of 30-day mortality and the impact of statin therapy on MINS is likely a dominant pathway explaining the associated lower risk of death with statins.¹²

Our results add important information to the available evidence. When compared with existing observational studies, VISION collected data prospectively, included a broader range of patients and types of surgeries in several countries, and actively monitored for outcomes. VISION also complements the findings from RCTs, since it observed a larger number of events and found effects of smaller magnitude, which are consistent with the benefits reported for long-term statin therapy. Additionally, VISION is the only study to report the effects of statins on MINS.

Our study's strengths include the large sample of patients from eight countries in five continents. We included a representative sample of patients undergoing non-cardiac surgery minimizing the risk of selection bias. All patients had the same troponin assay measured after surgery. A total of 99.7% of the patients completed the follow-up, and we had complete data on the 27 variables evaluated in our propensity score. Outcomes were centrally adjudicated. The model demonstrated good discrimination and calibration. Finally, the results were robust to the sensitivity analysis assumptions.

Our study has limitations. Even with adjustment for covariates, the use of statins in an observational dataset may represent a surrogate for unmeasured confounders that relate to prognosis. Despite matching, standardized differences remained >10% in five pre-operative variables. Yet all of these variables were more frequently seen in the Statin group compared with No Statin group in the matched population, thus making it very unlikely that they could have explained the benefit seen with statins. If optimal matching had been possible, then the results seen for statins users might have been even more pronounced. We did not collect data on type and dosing of statins used and could not evaluate whether the results varied based on these factors. We did not obtain data related to liver and muscle function. Nevertheless, it is unlikely that perioperative statin therapy leads to a higher risk for these events than with long-term statin use in routine practice.

In VISION, contraindications to statins were not recorded. Consequently, some control patients might have presented a contraindication to statins, which could introduce bias. However, absolute contraindications to statin use are rare, and this is unlikely to have influenced our results. Finally, in the main analysis, we considered any patients pre-operatively treated with a statin to have continued a statin after surgery. If a post-operative rebound effect exists with statin withdrawal and some patients taking a statin before surgery did not continue it after surgery, our assumption may have contributed to an underestimation of the treatment effect. Nevertheless, our sensitivity analysis, which included only patients who were continuously treated with statins, found similar results to the main analysis.

In summary, in a large cohort of patients undergoing non-cardiac surgery, pre-operative statin use was associated with a lower risk of 30 day cardiovascular events. Statins may represent a potentially beneficial intervention to prevent cardiovascular complications in this setting. Our results require confirmation in a large perioperative statin RCT.

Supplementary material

Supplementary material is available at European Heart Journal online.

Funding

Canada grants: Canadian Institutes of Health Research; Heart and Stroke Foundation of Ontario; Academic Health Science Centres Alternative Funding Plan Innovation Fund; McMaster University (CLARITY Group and Departments of Clinical Epidemiology and Biostatistics, PHRI, Cardiology, Surgery, Surgical Associates Research, Anesthesiology, Medicine); Hamilton Health Sciences (New Investigator Fund and Summer Studentships); Hamilton Health Sciences Grant; Ontario Ministry of Resource and Innovation Grant; Stryker Canada; Saint Joseph's Healthcare, Department of Medicine; Father Sean O'Sullivan Research Centre; Canadian Network and Centre for Trials Internationally; Winnipeg Health Sciences Foundation Operating Grant; University of Manitoba (University Medical Group and Departments of: Surgery, Surgery GFT Research, Faculty of Dentistry Operational Fund, Anesthesia); Diagnostic Services of Manitoba Research; Manitoba Medical Services Foundation Grant; Manitoba Health Research Council. S.S. was funded by the Rudy Falk Clinician Scientist Award. Australia: National Health and Medical Research Council Program; Australian and New Zealand College of Anesthesiologists. Brazil: PROADI-SUS – Ministry of Health, China: Public Policy Research Fund, Research Grant Council, Hong-Kong SAR. Colombia: School of Nursing, Universidad Industrial de Santander; Grupo de Cardiología Preventiva, Universidad Autónoma de Bucaramanga; Fundación Cardioinfantil – Instituto de Cardiología; Alianza Diagnóstica S.A. India: St. John's Medical College and Research Institute Grant, Division of Clinical Research and Training Grant. Malaysia: University of Malaya Research Grant; University of Malaya, Penyelidikan Jangka Pendek Grant. Spain: Instituto de Salud Carlos III, Fundació La Marató de TV3. The USA: American Heart Association Grant. The UK: National Institute for Health Research. Dr Nagele was funded by the National Institute for General Medical Sciences, NIH, and Washington University Institute of Clinical and Translational Sciences.

Conflict of interest: Funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation or approval of the article. Roche-Diagnostics provided TnT assays and financial support for the Study. O.B. has received grants from Astra-Zeneca, Bayer, Amgen, and Boehringer-Ingelheim. P.J.D. has received grants from Roche-Diagnostics and Abbott-Diagnostics. Other authors reported no conflict of interest.