Abstract
Aims
Prolonged QRS duration (QRSd) on the electrocardiogram (ECG) has been associated with cardiac structural and functional abnormalities by echocardiography and an increased risk of heart failure (HF). Data are sparse on these relationships in middle-aged and elderly individuals free of baseline cardiovascular disease with respect to cardiac magnetic resonance imaging (MRI). We sought to determine whether QRSd is associated with incident HF and measures of cardiac structure and function by cardiac MRI.
Methods and results
We analysed baseline ECGs in the Multi-Ethnic Study of Atherosclerosis (MESA) to determine whether QRSd >100 ms was associated with incident HF. We adjusted for demographic and clinical risk factors, as well as MRI measures of left ventricular (LV) structure and function. Among 4591 eligible participants (51% women; 39% white; mean age 61 years), 75 developed incident HF over a mean follow-up of 7.1 years. QRSd >100 ms was significantly associated with MRI measures of cardiac structure and function, as well as incident HF, even after adjustment for demographic covariates [hazard ratio (HR) 2.10, 95% confidence interval (CI) 1.29–3.42; P = 0.003] and clinical risk factors (HR 1.86, 95% CI 1.14–3.03; P = 0.01). With further adjustment for individual LV structural measures, findings were attenuated to non-significance. Separate adjustment for LV functional measures yielded only mild attenuation.
Conclusion
In middle-aged and older adults without cardiovascular disease, a QRSd >100 ms was significantly associated with incident HF. After adjustment for LV structural measures, the association was attenuated to non-significance, suggesting that prolonged QRSd is potentially a useful marker of LV structure that may predispose to HF risk.
Keywords: Electrocardiogram, Heart failure, Magnetic resonance imaging, QRS duration
Introduction
The QRS duration (QRSd) on the 12-lead electrocardiogram (ECG)represents conduction through the specialized cardiac conduction system and ventricular myocardium. A QRSd ≥ 120 ms has traditionally been classified with complete bundle branch blocks and unspecified intraventricular conduction delay, and a QRSd <100 ms has been considered ‘normal’.1 A QRSd between 100 and 120 ms is considered prolonged, specifying either incomplete bundle branch block or intraventricular conduction delay. The degree of QRS widening may be a manifestation of left ventricular (LV) structure (such as increased LV mass or LV dimension), but can also suggest functional abnormalities, such as LV systolic dysfunction (LVSD).
QRSd cut-off points of >100 and ≥ 120 ms have been associated with increased risk of heart failure (HF).2 Among middle-aged and elderly individuals free of cardiovascular disease, QRS prolongation ≥ 120 ms is uncommon, and data are sparse regarding whether any QRSd >100 ms is associated with incident HF. Similarly, it is unclear whether such prolongation represents cardiac structural or functional findings as assessed by cardiac magnetic resonance imaging (MRI). Therefore, in healthier populations, lesser degrees of QRSd prolongation may be more appropriate to study and may still confer HF risk.
Previous studies have demonstrated an association between QRSd and measures of cardiac structure and function, but were limited largely to Caucasians,2,3 and echocardiography was used, which is subject to limitations.4 Studies examining the associations between QRSd and incident HF and cardiac measures of LV structure and function have not been performed in a cohort free of baseline cardiovascular disease using cardiac MRI, a more precise imaging tool of the ventricular myocardium.5 We sought to identify whether a prolonged QRSd—defined a priori as >100 ms—is associated with incident HF over 7.1 years of follow-up in the Multi-Ethnic Study of Atherosclerosis (MESA), a well-phenotyped multiethnic, middle-aged, and older cohort. In addition, to explore potential mechanisms linking prolonged QRSd and incident HF, we sought to determine whether QRSd was associated with cardiac MRI measures of LV structure or function.
Methods
Study sample
Inclusion criteria and methods of the MESA study have been described previously.6 In brief, between July 2000 and August 2002, a total of 6814 men and women aged 45–84 years old and free of clinically apparent cardiovascular disease were recruited from six US communities: Baltimore City and Baltimore County, MD; Chicago, IL; Forsyth County, NC; Los Angeles County, CA; Northern Manhattan and the Bronx, NY; and St. Paul, MN. Consenting participants had an ECG performed at their baseline examination and a cardiac MRI scan a median of 16 days after the baseline evaluation; 95% of studies were completed by 11 weeks after the baseline examination. This study complies with the Declaration of Helsinki, and the institutional review boards at all participating centres approved the study; all participants gave informed consent.
We excluded participants with poor quality ECGs (n =48), antiarrhythmic medication use (n =35), Wolff–Parkinson–White syndrome or conduction system disturbance (n =10), incomplete ECG or MRI measurement at baseline (n =2067), no HF event information or incomplete baseline measurement (n =33), and interim myocardial infarction (n =30), leaving 4591 participants eligible for analysis (see Supplementary material, Figure S1).
Risk factor measures
Standardized questionnaires were used to obtain information about self-reported race/ethnicity, smoking history, and medication use for high blood pressure, high cholesterol, and diabetes. Smoking was defined by current use. Subjects had height and weight measured wearing light clothing and no shoes. Resting blood pressure was measured three times with participants in the seated position with a Dinamap model Pro 100 automated oscillometric sphygmomanometer (Critikon, GE Healthcare, Waukesha, WI, USA). The average of the last two measurements was used for analysis. Total cholesterol (TC), HDL cholesterol, and glucose levels were measured from blood samples obtained after a 12 h fast. LDL cholesterol was calculated with the Friedewald equation. Diabetes was defined as fasting glucose ≥126 mg/dL or use of hypoglycaemic medication. Body mass index (BMI; kg/m2) was calculated from weight measured to the nearest 0.5 kg and height to the nearest 0.1 cm.
Magnetic resonance imaging protocol
Images were acquired by 1.5 T MR scanners with determination of LV mass and geometry as previously described.7 All MRI studies were submitted to the core MESA MRI Reading Center at Johns Hopkins Hospital where all analyses were performed and quality assurance was maintained per study protocol.7–9
Electrocardiogram analysis of QRS duration
The study ECGs were recorded using MAC 1200 ECG machines (Marquette Electronics, Milwaukee, WI, USA) in all clinical centres. The ECGs were processed in a central laboratory at the EPICARE Center (Wake Forest University, Winston-Salem, NC, USA). All ECGs were visually inspected for technical errors and inadequate quality. The methodology for ECG analysis has been described elsewhere.10
Adjudication of events
Participants were followed for the development of incident cardiovascular events up to a mean of 7.1 years from their baseline examinations. In addition to three follow-up MESA study examinations, a telephone interviewer contacted each participant every 9–12 months to inquire about all interim hospital admissions, cardiovascular outpatient diagnoses, and deaths. Confirmation of self-reported diagnoses included requesting copies of all death certificates and medical records for all hospitalizations and outpatient cardiovascular diagnoses. Medical records were successfully obtained on an estimated 98% of hospitalized cardiovascular events and information on 95% of outpatient cardiovascular diagnostic encounters. Follow-up telephone interviews were completed in 92% of living participants.
Trained personnel abstracted any medical records suggesting possible cardiovascular events. Two physicians from the MESA study events committee, blinded to MESA study MRI and ECG results, independently reviewed all medical records for endpoint classification using pre-specified criteria. Event classification was determined by reviewers, and adjudication was performed as necessary.
Reviewers classified HF as definite or probable, as previously described.8 In brief, definite and probable HF required clinical symptoms (e.g. shortness of breath) or signs (e.g. oedema). Asymptomatic disease was not an endpoint. Probable HF required a physician diagnosis of HF and medical treatment for HF. For our analysis, we counted definite and probable HF as outcome events. Reviewers classified myocardial infarction as definite, probable, or absent, based on pre-specified criteria, including symptoms (e.g. chest pain), ECG abnormalities, and cardiac biomarker levels.
Statistical methods
Baseline characteristics were compared according to QRS prolongation status (>100 ms) using general linear models for continuous variables and χ2 tests for categorical variables. Pearson partial correlation coefficients were computed to assess associations among MRI measurements and continuous QRS duration measurement after adjustment for age and sex. To visualize the relationship between QRS prolongation and incidence of HF, a Kaplan–Meier plot was generated. We examined the association between QRS duration >100 ms and the risk of HF over a mean follow-up of 7.1 years using Cox proportional hazards regression. We verified that the assumption of proportionality of hazards was appropriate. Multivariable models using baseline characteristics were initially adjusted for demographic covariates of age, sex, and race, and then clinical covariates of systolic blood pressure, antihypertensive therapy, TC/HDL cholesterol ratio, BMI, diabetes, and current smoking status. Further adjustment for MRI measures of LV structure and function were performed with each MRI measure added separately to the model with demographic and clinical covariates. Specific structural measures studied were LV end-diastolic volume (LVEDV) index, LV end-systolic volume (LVESV) index, LV mass to volume ratio, and LV mass index (each in a separate analysis). Specific functional measures studied were LV stroke volume index and LV ejection fraction (in separate analyses). To investigate the dose–response relationship of QRS duration and HF risk, the penalized spline was implemented in the Cox model which can provide non-parametric estimates for the hazard ratio (HR) of QRS duration. All statistical analyses were performed using SAS statistical software v. 9.1 (Cary, NC, USA). The penalized spline model was implemented using coxph and pspline functions in R version 2.14.1. A two-tailed P-value <0.05 was considered statistically significant.
Results
Among 4591 eligible participants, 51.1% were women, 26.3% black, 22.2% Hispanic, and 12.5% Chinese, with a mean age of 61.3 years. Baseline characteristics for the study sample, stratified by QRS duration, are shown in Table 1. There were statistically significant differences in measured demographic and clinical characteristics between those with normal (≤100 ms) and prolonged (>100 ms) QRSd. Participants with QRSd >100 ms were more likely to be older men, have higher seated systolic and diastolic blood pressure, greater body BMI, larger TC/HDL cholesterol ratio, greater antihypertensive medication use, and lower HDL cholesterol. Table 2 shows characteristics of the study population, stratified by the presence of HF. Those with HF were more likely to be older men, with higher seated systolic blood pressure, greater BMI, more antihypertensive medication use, diabetes, and greater current smoking status. Table 3 shows the associations between QRSd and MRI characteristics of the study population and correlation after adjustment for age and sex. Those with QRSd >100 ms had significantly greater LVEDV index, LVESV index, LV mass index, LV mass/volume ratio, LV stroke volume index, and LV wall thickness (LVWT), but lower ejection fraction. There were statistically significant correlations between MRI measures and QRS duration, except the LV mass/volume ratio.
Table 1.
Characteristics of the study population, stratified by QRS duration at baseline
| Characteristic | QRSd ≤100 ms (n = 3691) | QRSd >100 ms (n = 900) | P-value |
|---|---|---|---|
| Age, years | 61.1 ± 10 | 62.2 ± 10.3 | <0.01 |
| Sex, female, (%) | 57.6 | 24.3 | <0.01 |
| Race (%) | <0.01 | ||
| Caucasians | 37.5 | 44.9 | |
| Black | 13.4 | 9.0 | |
| Hispanic | 26.6 | 25.3 | |
| Chinese | 22.5 | 20.8 | |
| Seated SBP (mmHg) | 124.8 ± 20.9 | 127.4 ± 21.5 | <0.01 |
| Seated DBP (mmHg) | 71.7 ± 10.3 | 73.4 ± 10.1 | <0.01 |
| BMI (kg/m2) | 27.7 ± 5.0 | 28.2 ± 4.5 | <0.01 |
| HDL (mg/dL) | 51.9 ± 15.2 | 47.8 ± 13.3 | <0.01 |
| TC/HDL cholesterol ratio | 4.0 ± 1.2 | 4.2 ± 1.2 | <0.01 |
| Treated HTN (%) | 31.0 | 36.4 | <0.01 |
| Diabetes (%) | 11.0 | 13.3 | 0.05 |
| Current smoker (%) | 12.8 | 13.1 | 0.81 |
BMI, body mass index; DBP, diastolic blood pressure; HTN, hypertension; QRSd, QRS duration; SBP, systolic blood pressure; TC, total cholesterol.
Table 2.
Characteristics of the study population, stratified by presence of heart failure
| Characteristic | Heart failure (n = 75) | No heart failure (n = 4516) | P-value |
|---|---|---|---|
| Age, years | 68.1 ± 8.5 | 61.2 ± 10.0 | <0.01 |
| Sex, female, (%) | 33.3 | 51.4 | <0.01 |
| Race (%) | 0.02 | ||
| Caucasians | 37.3 | 39.0 | |
| Black | 4.0 | 12.7 | |
| Hispanic | 40.0 | 26.1 | |
| Chinese | 18.7 | 22.2 | |
| Seated SBP (mmHg) | 137.6 ± 19.4 | 125.1 ± 21.1 | <0.01 |
| Seated DBP (mmHg) | 74.3 ± 11.1 | 72.0 ± 10.2 | 0.05 |
| BMI (kg/m2) | 29.4 ± 5.3 | 27.8 ± 4.9 | <0.01 |
| HDL (mg/dL) | 50.2 ± 15.5 | 51.1 ± 14.9 | 0.61 |
| TC/HDL cholesterol ratio | 4.00 ± 1.2 | 4.06 ± 1.2 | 0.69 |
| QRSd (ms) | 102.8 ± 20.9 | 93.3 ± 13.5 | <0.01 |
| Treated HTN (%) | 54.7 | 32.4 | <0.01 |
| Diabetes (%) | 28.0 | 11.2 | <0.01 |
| Current smoker (%) | 22.7 | 12.7 | 0.01 |
| QRS prolongation (%) | 38.7 | 19.3 | <0.01 |
BMI, body mass index; DBP, diastolic blood pressure; HTN, hypertension; QRSd, QRS duration; SBP, systolic blood pressure; TC, total cholesterol.
Table 3.
Magnetic resonance imaging measures of the study population by QRS status and correlation after adjustment for age and sex
| Mean and 95% CI of MRI measures by QRS status |
Partial Pearson's correlation coefficients between MRI measure and QRSd |
||||
|---|---|---|---|---|---|
| QRSd ≤100 ms (n = 3691) | QRSd >100 ms (n = 900) | P-value | r | P-value | |
| LVEDV (mL) | 67.5 (67.0–67.9) | 72.3 (71.4–73.2) | <0.01 | 0.20 | <0.01 |
| LVESV (mL) | 21.0 (20.8–213) | 23.8 (23.3–24.3) | <0.01 | 0.19 | <0.01 |
| LV mass (mL) | 77.0 (76.5–7.4) | 83.8 (82.9–84.8) | <0.01 | 0.22 | <0.01 |
| LV mass/volume ratio | 0.91 (0.90–0.91) | 0.92 (0.91–0.93) | 0.04 | 0.02 | 0.30 |
| LV basal wall thickness (mm) | 9.9 (9.9–10.0) | 10.4 (10.3–10.5) | <0.01 | 0.12 | <0.01 |
| LV mid-wall thickness (mm) | 9.2 (9.2–9.3) | 9.7 (9.6–9.8) | <0.01 | 0.12 | <0.01 |
| LVSV (mL) | 46.4 (46.1–6.7) | 48.5 (47.9–49.1) | <0.01 | 0.11 | <0.01 |
| LVEF (%) | 69.2 (68.9–69.4) | 67.8 (67.3–68.2) | <0.01 | −0.11 | <0.01 |
Analyses were adjusted for age and sex. Mean and 95% confidence interval (CI) values are presented.
All MRI characteristics except wall thickness, mass/volume ratio, and LVEF were indexed to body surface area.
CI, confidence interval; LV, left ventricular LVEDV, left ventricular end-diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end-systolic volume; LVSV, left ventricular stroke volume; MRI, magnetic resonance imaging; QRSd, QRS duration.
Incident heart failure
Seventy-five participants developed incident HF over a mean follow-up of 7.1 years. Compared with participants with normal QRSd, those with QRSd >100 ms had significantly greater risk for incident HF, with an unadjusted HR of 2.64 [95% confidence interval (CI) 1.66–4.20, P < 0.0001; Table 4]. The mean QRSd for the 75 participants who developed HF was 102.8 ± 20.9 ms; the mean QRSd for the participants who did not develop HF was 93.3 ± 13.5 ms. The mean ejection fraction for all participants was 68.9 ± 7.2%, and for the 75 participants who developed HF it was 60.8 ± 12.6% (P < 0.01). Figure 1 shows the Kaplan–Meier plot indicating that patients with a QRSd >100 ms were significantly more likely to develop incident HF than those with a QRSd ≤100 ms. The sensitivity and specificity of a QRSd >100 ms predicting HF was 38.7% and 80.7%, respectively. The positive and negative predictive values were 3.2% and 98.8%.
Table 4.
Hazard ratios for incident heart failure associated with QRS duration >100 ms
| Adjustment | HR (95% CI) for association of QRSd >100 ms with incident heart failure | P-value for association of QRSd >100 ms with incident heart failure |
|---|---|---|
| Unadjusted | 2.64 (1.66–4.20) | <0.01 |
| Age, sex, race (demographics) | 2.10 (1.29–3.42) | <0.01 |
| Clinical + demographics | 1.86 (1.14–3.03) | 0.01 |
| Clinical + demographics + LVEDVI | 1.26 (0.76–2.11) | 0.37 |
| Clinical + demographics + LVESVI | 1.24 (0.74–2.08) | 0.41 |
| Clinical + demographics + LVMI | 1.28 (0.76–2.16) | 0.35 |
| Clinical + demographics + LV mass/volume ratio | 1.86 (1.14–3.03) | 0.01 |
| Clinical + demographics + SVI | 1.79 (1.09–2.93) | 0.02 |
| Clinical + demographics + LVEF | 1.55 (0.94–2.56) | 0.09 |
Clinical covariates adjusted for in the analysis included systolic blood pressure, antihypertensive therapy, total cholesterol/HDL cholesterol ratio, body mass index; diabetes, current smoking; cardiac magnetic resonance imaging parameters included left ventricular end-diastolic diameter indexed to body surface area (LVEDVI); left ventricular end-systolic diameter indexed to body surface area (LVESVI); left ventricular mass indexed to body surface area (LVMI); stroke volume indexed to body surface area (SVI); and left ventricular ejection fraction (LVEF).
CI, confidence interval; HR, hazard ratio; QRSd, QRS duration.
Figure 1.
Probability of survival free of heart failure by baseline QRS duration. The graphic depicts survival free of heart failure by baseline QRS duration.
Results from multivariable-adjusted models are shown in Table 4. The association between QRSd >100 ms and incident HF was modestly attenuated but remained significant after adjustment for age, sex, and race, and after further adjustment for clinical covariates. When MRI measures of LV structure were added individually to the model with demographic and clinical covariates, the association between QRSd >100 ms and incident HF was further attenuated to non-significance. However, QRSd >100 ms was still associated with incident HF after adjustment for LV mass/volume ratio, a parameter used to describe concentric remodelling.8 However, when MRI measures of LV function were added, separately, to the model with demographic and clinical covariates, QRSd >100 ms was still associated with incident HF, without substantial attenuation beyond that seen with clinical and demographic covariates. When we examined the association between QRSd as a continuous variable with incident HF, we found a similar pattern of results (Supplementary material, Table S1).
Secondary analyses
There were no significant interactions by sex–race/ethnicity group in the association of QRSd with incident HF or with MRI parameters (Table 3); however, some subgroups had few events. To assess the gradient of HF risk by QRSd, we also examined the risk for incident HF associated with QRSd between 101 and 119 ms (n =704), excluding those with QRSd ≥120 ms (n =196). In this analysis, QRSd between 101 and 119 ms, compared with QRS ≤100 ms, was associated with incident HF (unadjusted HR 1.85, 95% CI 1.05–3.28). We observed a significant association of continuous QRSd and incident HF as well (Appendix 2). In order to examine the association of QRSd with incident HF across all values, we performed penalized spline models (Figure 2). In general, we observed a linear association between QRSd and HF incidence; of note, the risk for HF appears to be increased above a QRSd threshold of 100 ms, which coincides with our a priori cut-off point to define QRSd prolongation on clinical grounds. After further multivariable adjustment, the association of QRSd 101–119 ms with HF was attenuated substantially to non-significance.
Figure 2.
Penalized spline analysis of baseline QRS duration and risk of heart failure. The graphic depicts penalized spline analysis (solid line) with 95% confidence intervals (dashed lines). Risk for incident heart failure is at increased levels above a QRS duration of 100 ms.
Discussion
In this multiethnic cohort free of clinically manifest cardiovascular disease at baseline, we observed that the presence of a QRSd >100 ms is significantly associated with incident HF over a mean follow-up of 7.1 years, and there was a continuous, graded association between QRSd and incident HF risk. We also found that both structural and functional LV measures from MRI are significantly associated with QRSd. Given the attenuation in the association between QRSd and incident HF that we noted after adjustment for LV structural (but not LV functional) characteristics, our analysis suggests that QRSd is potentially a useful marker of LV structural measures and subsequent risk of HF in patients in this middle-aged and older cohort. Our data support the hypothesis that QRSd prolongation >100 ms may be a marker of LV structural abnormalities that predispose to incident HF.
Electrocardiographic findings associated with heart failure
QRSd has been shown to be a predictor of HF in a number of selected patient populations, including those with hypertension and ECG evidence of LV hypertrophy.11 Investigators found that participants who had persistence of or who developed a new QRSd ≥110 ms while undergoing treatment for hypertension were twice as likely to develop HF, independent of blood pressure-lowering, treatment modality, or regression of ECG evidence of LV hypertrophy. Our study extends the association of QRSd as a predictor of HF to a diverse, community-based sample without documented baseline cardiovascular disease.
Electrocardiographic correlates to left ventricular structural and functional measures
Framingham investigators found that QRSd is associated with echocardiographically measured LV end-diastolic dimension, LV mass, and fractional shortening.3 However, these studies were limited by including predominantly Caucasian participants and using echocardiographic measures, which may be subject to limitations.4 Our study benefited from a well-phenotyped cohort of participants with MRI-derived measures of LV structure and function, and broadens the findings of the Framingham investigators to an ethnically diverse group of participants. A prior study in the MESA cohort reported that MRI measures of cardiac structure and function were associated with subsequent cardiovascular events,8 such as HF, but only LV mass in the most extreme category (>95th percentile) was associated with HF risk. Of note, these investigators reported that participants in whom HF events developed were more likely to have LV hypertrophy, which was predicted by extreme values in LV mass, and were more likely to be seen in African-American participants. Our study found that QRSd was strongly associated with cardiac structural parameters, including LV mass, LVWT, and chamber size as assessed by cardiac MRI, not just LV mass alone, further supporting the hypothesis that QRSd may be a marker of LV structure in this cohort. We did not find a race/ethnicity association by QRSd and HF risk in our study. Further research is needed to explore these potential differences.
Potential associations between QRS duration and left ventricular structure and function
Historically, a prolonged QRSd (≥120 ms) has been considered ‘abnormal’, and has been associated with adverse cardiovascular outcomes. Patients with marked prolongation of the QRSd are more likely to have systolic dysfunction due to factors such as myocardial scarring from prior myocardial infarction and dilated ventricular chambers.12–15 However, lesser degrees of QRSd prolongation may also be markers of HF risk, particularly in patient groups free of baseline cardiovascular disease. Investigators recently reported that cardiac structure and function may contribute to the higher incidence of HF observed in older adults, particularly HF that presents with a normal ejection fraction.16,17 This study describes a ‘natural progression’ in both sexes of ventricular remodelling, which is characterized by steady increases in LVWT, followed by a decrease in LV diastolic dysfunction (LVDD) and LVSD, and a progressive increase in fractional shortening. These changes result in an overall decrease in LV cavity size concurrent with increases in LVWT. Thus, smaller LV volumes in older individuals may contribute to the lower haemodynamic tolerance for a preload challenge contributing to HF incidence.18
Our findings appear consistent with these investigators, with some notable differences. First, thicker left ventricular walls (and perhaps greater LV mass) appear to be consistently associated with a prolonged QRSd and HF risk. Thicker ventricular walls may contribute to delayed cardiac conduction due to properties such as anisotropy and fibre orientation, which could prolong impulse duration. Therefore, QRSd may represent a reliable marker of LV hypertrophy and greater LV mass. Secondly, larger, not smaller, chamber dimensions, appear to be associated with prolonged QRSd and HF risk. The characteristics of larger LVEDV and LVESV and increased LV mass appear to be more significantly associated with participants in our study with a QRS >100 ms. Thirdly, due to thicker LV walls (and perhaps mass) and smaller chamber dimensions, one may expect both ejection fraction and stroke volume to be higher (not lower). Our results suggest that participants with a QRS >100 ms and HF were more likely to have a slightly lower (yet preserved) ejection fraction and stroke volume compared with those participants with QRSd ≤100 ms.
Current study in context
While sophisticated, multivariable algorithms may be useful in predicting HF, these often require several pieces of clinical information, some of which are not readily available to the practising clinician, or, as in the case of MRI, not indicated or unavailable at some institutions. ECGs are widely available in clinical practice at low cost, and, therefore, clinicians can use this tool to assist in understanding the risk for HF in patients within this age group. Given that QRSd may be an important additional marker of HF risk in our study, future studies should examine the predictive utility of QRSd in patient populations with more HF events.
Differences in MRI characteristics between those developing and not developing HF based on QRSd are small, with a difference of 5 mL in LVEDV, 3 mL in LVESV, 7 g in LV mass, 0.5 cm in wall thickness, and 1.4% in ejection fraction. On balance, the cumulative structural changes, rather than a single metric, may reflect a greater QRSd and, therefore, risk of HF in this cohort. It is important to point out that functional measures alone (ejection fraction and stroke volume) did not attenuate the relationship between QRSd and incident HF. On the basis of our data, we cannot universally advocate use of the ECG as a predictor of HF. The strong negative predictive value of 98.8% underscores the likelihood that a QRSd ≤100 ms in this cohort is not associated with HF. However, we believe our data do suggest a strong association between QRSd and cardiac structural parameters as detected by cardiac MRI. Longitudinal studies assessing incremental changes in QRSd and LV structure and function in this cohort will better address this hypothesis.
Strengths and weaknesses of the current study
There are several strengths of the current study. First, we used standard definitions for HF events which were adjudicated by trained physicians, high-quality digital ECG phenotyping, and MRI assessment of LV structural and functional measures to strengthen the internal validity of the findings. Secondly, the MESA cohort is a well-phenotyped, diverse, multiethnic study population, and included a large proportion of women, which strengthens the external validity of the study.
Despite these strengths, there are several weaknesses that should be considered. For example, we do not have longitudinal assessment of QRSd and LV structural or functional changes to assess temporality. In addition, our analysis did not account for downstream treatment effects which may have modulated both cardiac conduction and LV structure and function. We also do not have information related to myocardial scarring from unknown (or unmeasured) processes (such as silent myocardial infarction), since MRIs were not uniformly performed with contrast media to assess delayed enhancement. Myocardial scarring is another potential mechanism that could alter the conduction system and prolong the QRSd and increase risk for HF. However, baseline ECGs and detailed questionnaires to screen for interim events (including interim myocardial infarction) were performed (and excluded from our analysis) to avoid such potential confounding. Finally, we did not have enough clinical events for robust analysis by sex–race/ethnicity groups.
Conclusion
We observed that the presence of a QRSd >100 ms is significantly associated with incident HF, and there was a continuous, graded association between QRSd and incident HF risk. We also found that LV structural measures from MRI are significantly associated with QRSd, suggesting a potential mechanistic link between structural MRI findings and QRSd and increased risk of incident HF. Whether QRSd, a simple and readily obtainable finding on the ECG, can be used to predict prospective HF events needs further study.
Supplementary material
Supplementary material is available at European Journal of Heart Failure online.
Funding
The National Heart, Lung, and Blood Institute (contracts N01-HC-95159 through N01-HC-95169).
Conflict of interest: none declared.
Supplementary Material
Acknowledgements
The authors thank other investigators, staff, and participants of the MESA study for their valuable contributions. A full list of participating MESA investigators and institutions can be found at http://www.mesa-nhlbi.org.
References
- 1.Josephson ME. Intraventricular conduction disturbances. In: Josephson ME, editor. Clinical Cardiac Electrophysiology: Techniques and Interpretations. 3rd Ed. Philadelphia: Lippincott Williams & Wilkins; 2002. pp. p110–139. [Google Scholar]
- 2.Dhingra R, Pencina MJ, Wang TJ, Nam BH, Benjamin EJ, Levy D, Larson MG, Kannel WB, D'Agostino RB, Sr, Vasan RS. Electrocardiographic QRS duration and the risk of congestive heart failure: the Framingham Heart Study. Hypertension. 2006;47:861–867. doi: 10.1161/01.HYP.0000217141.20163.23. [DOI] [PubMed] [Google Scholar]
- 3.Dhingra R, Nam BH, Benjamin EJ, Wang TJ, Larson MG, D'Agostino RB, Sr, Levy D, Vasan RS. Cross-sectional relations of electrocardiographic QRS duration to left ventricular dimensions: the Framingham Heart Study. J Am Coll Cardiol. 2005;45:685–689. doi: 10.1016/j.jacc.2004.11.046. [DOI] [PubMed] [Google Scholar]
- 4.Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, Picard MH, Roman MJ, Seward J, Shanewise JS, Solomon SD, Spencer KT, Sutton MS, Stewart WJ Chamber Quantification Writing Group: American Society of Echocardiography's Guidelines and Standards Committee: European Society of Echocardiography. Recommendations for chamber quantification: a report from the American Society of Echocardiography's Guidelines and Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr. 2005;18:1440–1463. doi: 10.1016/j.echo.2005.10.005. [DOI] [PubMed] [Google Scholar]
- 5.Pohost GM, Hung L, Doyle M. Clinical use of cardiovascular magnetic resonance. Circulation. 2003;108:647–653. doi: 10.1161/01.CIR.0000083233.86078.3E. [DOI] [PubMed] [Google Scholar]
- 6.Bild DE, Bluemke DA, Burke GL, Detrano R, Diez Roux AV, Folsom AR, Greenland P, Jacob DR, Jr, Kronmal R, Liu K, Nelson JC, O'Leary D, Saad MF, Shea S, Szklo M, Tracy RP. Multi-Ethnic Study of Atherosclerosis: objectives and design. Am J Epidemiol. 2002;156:871–881. doi: 10.1093/aje/kwf113. [DOI] [PubMed] [Google Scholar]
- 7.Natori S, Lai S, Finn JP, Gomes AS, Hundley WG, Jerosch-Herold M, Pearson G, Sinha S, Arai A, Lima JA, Bluemke DA. Cardiovascular function in multi-ethnic study of atherosclerosis: normal values by age, sex, and ethnicity. Am J Roentgenol. 2006;186:S357–S365. doi: 10.2214/AJR.04.1868. [DOI] [PubMed] [Google Scholar]
- 8.Bluemke DA, Kronmal RA, Lima JAC, Liu K, Olson J, Burke GL, Folsom AR. The relationship of left ventricular mass and geometry to incident cardiovascular events: the MESA (Multi-Ethnic Study of Atherosclerosis) study. J Am Coll Cardiol. 2008;52:2148–2155. doi: 10.1016/j.jacc.2008.09.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Rodriguez CJ, Diez-Roux AV, Moran A, Jin Z, Kronmal RA, Lima J, Homma S, Bluemke DA, Barr RG. Left ventricular mass and ventricular remodeling among Hispanic subgroups compared with non-Hispanic blacks and whites. MESA (Multi-Ethnic Study of Atherosclerosis) J Am Coll Cardiol. 2010;55:234–242. doi: 10.1016/j.jacc.2009.08.046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ilkhanoff L, Soliman EZ, Ning H, Liu K, Lloyd-Jones DM. Factors associated with development of prolonged QRS duration over 20 years in healthy young adults: the Coronary Artery Risk Development in Young Adults study. J Electrocardiol. 2012;45:178–184. doi: 10.1016/j.jelectrocard.2011.11.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Okin PM, Devereux RB, Kjeldsen SE, Edelman JM, Dahlof B. Incidence of heart failure in relation to QRS duration during antihypertensive therapy: the LIFE study. J Hypertens. 2009;27:2271–2277. doi: 10.1097/HJH.0b013e328330b66b. [DOI] [PubMed] [Google Scholar]
- 12.Iuliano S, Fisher SG, Karasik PE, Fletcher RD, Singh SN. Department of Veterans Affairs Survival Trial of Antiarrhythmic Therapy in Congestive Heart Failure. Am Heart J. 2002;143:1085–1091. doi: 10.1067/mhj.2002.122516. QRS duration and mortality in patients with congestive heart failure. [DOI] [PubMed] [Google Scholar]
- 13.Shenkman HJ, Pampati V, Khandelwal AK, McKinnon J, Nori D, Kaatz S, Sandberg KR, McCullough PA. Congestive heart failure and QRS duration: establishing prognosis study. Chest. 2002;122:528–534. doi: 10.1378/chest.122.2.528. [DOI] [PubMed] [Google Scholar]
- 14.Shamim W, Yousufuddin M, Cicoria M, Gibson DG, Coats AJ, Henein MY. Incremental changes in QRS duration in serial electrocardiograms over time identify high risk elderly patients with heart failure. Heart. 2000;88:47–51. doi: 10.1136/heart.88.1.47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Desai AD, Yaw TS, Yamazaki T, Kaykha A, Chun S, Froelicher VF. Prognostic significance of quantitative QRS duration. Am J Med. 2006;119:600–606. doi: 10.1016/j.amjmed.2005.08.028. [DOI] [PubMed] [Google Scholar]
- 16.Cheng S, Xanthakis V, Sullivan LM, Lieb W, Massaro J, Aragam J, Benjamin EJ, Vasan RS. Correlates of echocardiographic indices of cardiac remodeling over the adult life course: longitudinal observations from the Framingham Heart Study. Circulation. 2010;122:570–578. doi: 10.1161/CIRCULATIONAHA.110.937821. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Redfield MM, Jacobsen SJ, Borlaug BA, Rodeheffer RJ, Kass DA. Age and gender-related ventricular-stiffening: a community-based study. Circulation. 2005;112:2254–2262. doi: 10.1161/CIRCULATIONAHA.105.541078. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.