Sudden cardiac death (SCD) is a major cause of mortality in adults with congenital heart disease (CHD). Several risk factors for SCD including conduction disturbances and ventricular dysfunction have been described previously. However, electrocardiogram (ECG) and echocardiographic parameters may change over time, and the predictive value of such temporal changes, rather than their point estimates, for SCD remains unknown.
This was a retrospective case–control study in adults with CHD and proven or presumed SCD and matched controls. Data were obtained from three databases including 25 000 adults with CHD. Sequential measurements were performed on electrocardiograms and echocardiograms. Ventricular function was assessed by echocardiography and graded on a four-point ordinal scale: 1, normal [ejection fraction (EF) ≥50%]; 2, mildly impaired (EF 40–49%); 3, moderately impaired (EF 30–39%); and 4, severely impaired (EF < 30%). Overall, 131 SCDs (mean age 36 ± 14 years, 67% male) and 260 controls (mean age 37 ± 13 years, 63% male) were included. At baseline, median QRS duration was 108 ms (range 58–168 ms) in SCDs and 97 ms (range 50–168 ms) in controls and increased over time at a rate of 1.6 ± 0.5 vs. 0.5 ± 0.2 ms/year in SCDs and controls, respectively (P = 0.011). QT dispersion at baseline was 61 ms (range 31–168 ms) in SCDs and 50 ms (range 21–129 ms) in controls. QT dispersion increased at a rate of 1.1 ± 0.4 ms/year in SCD victims and decreased at a rate of 0.2 ± 0.2 ms/year in controls (P = 0.004). Increase of QRS duration ≥5 ms/year was associated with an increased risk of SCD [OR 1.9, 95% confidence interval (CI) 1.1–3.3, P = 0.013]. Change from any baseline systemic ventricular function (normal, mild, or moderately impaired) to severe ventricular dysfunction over time was associated with the highest risk of SCD (OR 16.9, 95% CI 1.8–120.1, P = 0.008).
In adults with CHD, QRS duration and ventricular dysfunction progress over time. Progression of QRS duration and the rate of impairment of ventricular function served to identify those at increased risk of SCD.
Sudden cardiac death (SCD) is a major cause of mortality in adults with congenital heart disease (CHD).
Several risk factors for SCD including conduction disturbances and ventricular dysfunction have been described previously.
However, electrocardiogram and echocardiographic parameters may change over time and the predictive value of such temporal changes, rather than their point estimates, for SCD are unknown.
QRS duration and ventricular dysfunction progress over time. Progression of QRS duration and the rate of impairment of ventricular function serve to identify adults with CHD who are at increased risk of SCD.
Introduction
Sudden cardiac death (SCD) is a major cause of mortality in adults with congenital heart disease (CHD).1–4 Potential risk factors for SCD in adults with CHD include functional status, prior arrhythmias, surgical characteristics, haemodynamic and electrocardiographic parameters. Measurements of these parameters at a single time point are valuable in predicting SCD.5–13 However, these parameters are known to vary considerably during the course of disease and treatment. Changes in these parameters over time rather than their absolute value may have prognostic importance. In patients with acquired heart disease and heart failure, increase in QRS duration over time has been associated with increased mortality, emphasizing that death may be predicted by looking carefully at the nature of changes in several clinical parameters rather than at their absolute value.14 Insight into time-dependent changes in functional parameters is relevant as it may serve to identify patients at risk for this devastating event. We have recently published a case–control that showed that several clinical parameters were associated with SCD in adults with CHD. These clinical parameters included documented prior supraventricular tachycardias, increased QRS duration, prolonged QT dispersion, and moderately to severely impaired systolic function of the systemic and/or subpulmonary ventricle.5
The purpose of the current study was to investigate whether changes over time in electrocardiographic and echocardiographic parameters may enhance prediction of SCD. Our hypothesis was that patients with SCD could be discriminated from controls by the course of these changed parameters, long before the occurrence of SCD.
Methods
Study population
Cases and controls were recruited from the CONCOR (CONgenital CORvitia) database, the Toronto Congenital Cardiac Centre for Adults, and the University Hospital Leuven. These three large databases contained >25 000 adults with CHD. A case–control study on adults with CHD and SCD has been published previously.5 Patients who died suddenly due to an arrhythmia were included (cases). For each case, two controls matched on CHD diagnosis, type of surgical intervention, date of surgical intervention, age, and gender were included. Control subjects were randomly selected from a list of potential candidates fulfilling the matching criteria. Clinical data of these subjects were collected afterwards. In this study, all patients with SCD and controls with more than one electrocardiogram (ECG) or echocardiogram were included. At least two sequential ECGs and/or echocardiograms were available in 131 out of the 171 SCD cases (77%) and 260 out of the 310 matched controls (84%).
Definitions
Sudden cardiac death is historically defined as (i) death due to cardiovascular causes within 1 h of onset or significant worsening of the symptoms or (ii) unwitnessed death during sleep. In this study, SCD included (i) proven or documented arrhythmic death [instantaneous death with documented ventricular fibrillation (VF) or ventricular tachycardia (VT)], (ii) arrhythmic death by exclusion (instantaneous death or circumstances compatible with SCD, without severe disease that would lead to death soon and in the absence of a non-arrhythmic cause of death at autopsy), and (iii) arrhythmic death by default (abrupt loss of consciousness and disappearance of pulse but no further data).15
Electrocardiogram measurements
Rhythm RR interval, QRS duration, QT interval, and QT dispersion were analysed from all leads of the standard (25 mm/s and 1 mV/cm) simultaneous 12-lead ECGs. Electrocardiograms were saved and analysed with ImageJ software (National Institutes of Health, Bethesda, MD, USA, http://rsb.info.nih.gov/ij/). All ECGs were digitized at 400 dpi. After four times enlargement of the digitized ECGs, calibrated measurements were performed on-screen by placing measurement points manually. For the QRS duration mean value was used, the QT dispersion was the maximum QT interval minus minimum QT interval. Investigators examining the ECGs were blinded to case status. Ventricular paced rhythms were excluded from analysis. Investigators examining the ECGs were blinded to case status. Complete left bundle branch block (LBBB) was defined as QRS duration ≥140 ms (men) or 130 ms (women), QS or rS in leads V1 and V2, and mid-QRS notching or slurring in ≥2 of leads V1, V2, V5, V6, I, and aVL.16 Complete right bundle branch block (RBBB) was defined as QRS duration ≥120 ms, qRS or qrS in I and V6, rsR′ or qR in V1, and ST depression and T inversion in right precordial and limb leads with ST-T vectors discordant to terminal mean QRS spatial vector.
Echocardiograms
Since in CHD the aorta does not always arise from the morphological left ventricle, we classified the ventricles as systemic and subpulmonary throughout the manuscript instead of left and right ventricles. Ventricular function was assessed qualitatively by 2D echocardiography and was graded in a four-point ordinal scale: 1, normal [ejection fraction (EF) ≥50%]; 2, mildly impaired (EF 40–49%); 3, moderately impaired (EF 30–39%); and 4, severely impaired (EF < 30%). Changes of ventricular function over time/per year were defined as deterioration or improvement within this four-point scale. For example, if systemic ventricular function (SVF) impairs by at least one point, this means that SVF deteriorates from mildly (2) to moderately (3) or moderately (3) to severely (4) or mildly (2) to severely (4).
Temporal analysis of data
All data were plotted backward in time and organized in the following time frames: within 1 year of last follow-up or SCD (t 0), 1–2 years (t-1), 3–4 years (t-2), 5–6 years (t-3), 7–8 years (t-4), and ≥9 years (t-5) from last follow-up.
Statistical analysis
Data analyses were performed with SPSS software for Windows (19.0 for Windows; SPSS, Inc., Chicago, IL, USA). For all analyses, two-tailed probability values <0.05 were considered statistically significant. Descriptive statistics for nominal data were expressed in absolute numbers and percentages. After confirming normality, mean values and standard deviation were calculated for normally distributed continuous variables. When comparing frequencies and means, the χ2-test and Student's t-test were used, respectively.
Changes in electrocardiographic and echocardiographic data over time were investigated using mixed linear models, taking into account sequential data of each patient (see Supplementary material online, Figures). To identify changes over time which may be associated with SCD, logistic regression analysis was used. Results were reported in odds ratios (ORs) with 95% confidence intervals (CIs).
Results
Characteristics of the study population
In the three databases including >25 000 adults with CHD, 171 patients died suddenly and were matched to 310 living controls.5 Hundred and thirty-one out of 171 SCD cases (77%) and 260 out of 310 matched controls (84%) had more than 1 ECG and/or transthoracic echocardiogram. For two SCDs, we were able to find only one control patient. These patients were included in this study. Clinical characteristics including the main cardiac diagnosis for SCD cases and controls are summarized in Table 1. The mean follow-up duration was 7 ± 5 years.
Characteristics of SCD cases and controls with sequential ECGs or echocardiograms
| Variable | Cases (n = 131) | Controls (n = 260) |
|---|---|---|
| Male, n (%) | 88 (67) | 165 (63) |
| Mean age at last follow-up | 36 ± 14 | 37 ± 13 |
| Main cardiac diagnosis, n (%) | ||
| Cyanotic Eisenmenger syndrome | 29 (22) | 44 (17) |
| (cc) TGA | 21 (16) | 50 (19) |
| Repaired TOF | 20 (15) | 40 (15) |
| Left-sided outflow lesions | 17 (13) | 35 (13) |
| Closed septal defects | 15 (11) | 26 (10) |
| VSD | 7 (5) | 12 (5) |
| ASD | 6 (5) | 11 (4) |
| AVSD | 2 (2) | 3 (1) |
| Cyanotic non-Eisenmenger | 15 (11) | 29 (11) |
| Fontan circulation | 6 (5) | 21 (8) |
| Ebstein anomaly | 3 (2) | 9 (3) |
| Other | 3 (2) | 2 (1) |
| Sequential electrocardiogram, n (%) | 120 (92) | 242 (93) |
| Sequential echocardiogram, n (%) | 81 (62) | 201 (77) |
| Variable | Cases (n = 131) | Controls (n = 260) |
|---|---|---|
| Male, n (%) | 88 (67) | 165 (63) |
| Mean age at last follow-up | 36 ± 14 | 37 ± 13 |
| Main cardiac diagnosis, n (%) | ||
| Cyanotic Eisenmenger syndrome | 29 (22) | 44 (17) |
| (cc) TGA | 21 (16) | 50 (19) |
| Repaired TOF | 20 (15) | 40 (15) |
| Left-sided outflow lesions | 17 (13) | 35 (13) |
| Closed septal defects | 15 (11) | 26 (10) |
| VSD | 7 (5) | 12 (5) |
| ASD | 6 (5) | 11 (4) |
| AVSD | 2 (2) | 3 (1) |
| Cyanotic non-Eisenmenger | 15 (11) | 29 (11) |
| Fontan circulation | 6 (5) | 21 (8) |
| Ebstein anomaly | 3 (2) | 9 (3) |
| Other | 3 (2) | 2 (1) |
| Sequential electrocardiogram, n (%) | 120 (92) | 242 (93) |
| Sequential echocardiogram, n (%) | 81 (62) | 201 (77) |
(cc) TGA, (congenitally corrected) transposition of great arteries; TOF, tetralogy of Fallot; VSD, ventricular septal defect; ASD, atrial septal defect; AVSD, atrioventricular septal defect.
Characteristics of SCD cases and controls with sequential ECGs or echocardiograms
| Variable | Cases (n = 131) | Controls (n = 260) |
|---|---|---|
| Male, n (%) | 88 (67) | 165 (63) |
| Mean age at last follow-up | 36 ± 14 | 37 ± 13 |
| Main cardiac diagnosis, n (%) | ||
| Cyanotic Eisenmenger syndrome | 29 (22) | 44 (17) |
| (cc) TGA | 21 (16) | 50 (19) |
| Repaired TOF | 20 (15) | 40 (15) |
| Left-sided outflow lesions | 17 (13) | 35 (13) |
| Closed septal defects | 15 (11) | 26 (10) |
| VSD | 7 (5) | 12 (5) |
| ASD | 6 (5) | 11 (4) |
| AVSD | 2 (2) | 3 (1) |
| Cyanotic non-Eisenmenger | 15 (11) | 29 (11) |
| Fontan circulation | 6 (5) | 21 (8) |
| Ebstein anomaly | 3 (2) | 9 (3) |
| Other | 3 (2) | 2 (1) |
| Sequential electrocardiogram, n (%) | 120 (92) | 242 (93) |
| Sequential echocardiogram, n (%) | 81 (62) | 201 (77) |
| Variable | Cases (n = 131) | Controls (n = 260) |
|---|---|---|
| Male, n (%) | 88 (67) | 165 (63) |
| Mean age at last follow-up | 36 ± 14 | 37 ± 13 |
| Main cardiac diagnosis, n (%) | ||
| Cyanotic Eisenmenger syndrome | 29 (22) | 44 (17) |
| (cc) TGA | 21 (16) | 50 (19) |
| Repaired TOF | 20 (15) | 40 (15) |
| Left-sided outflow lesions | 17 (13) | 35 (13) |
| Closed septal defects | 15 (11) | 26 (10) |
| VSD | 7 (5) | 12 (5) |
| ASD | 6 (5) | 11 (4) |
| AVSD | 2 (2) | 3 (1) |
| Cyanotic non-Eisenmenger | 15 (11) | 29 (11) |
| Fontan circulation | 6 (5) | 21 (8) |
| Ebstein anomaly | 3 (2) | 9 (3) |
| Other | 3 (2) | 2 (1) |
| Sequential electrocardiogram, n (%) | 120 (92) | 242 (93) |
| Sequential echocardiogram, n (%) | 81 (62) | 201 (77) |
(cc) TGA, (congenitally corrected) transposition of great arteries; TOF, tetralogy of Fallot; VSD, ventricular septal defect; ASD, atrial septal defect; AVSD, atrioventricular septal defect.
Electrocardiographic parameters
(A) Changes in mean QRS duration (ventricular pacing excluded) over years in SCDs and controls in serial ECG. Ventricular paced rhythms were present in 25 SCDs and 29 controls and excluded from analysis. (B) Changes in mean QT dispersion (ventricular pacing excluded) over years in SCDs and controls in serial ECGs.
At the last ECG available, median QRS duration was 122 ms (range 79–251 ms) vs. 108 ms (range 57–197 ms) (P < 0.001) and median QT dispersion 71 ms (range 17–193 ms) vs. 54 ms (range 13–136 ms, P < 0.001) for the SCD cases and controls, respectively.
The association of changes in QRS duration per year with the risk of SCD.
Increase in PR interval was 0.4 ± 0.6 ms/year in SCD cases and 0.2 ± 0.4 ms/year in controls (P = 0.736). QTc interval increased with 1.2 ± 0.6 and 0.3 ± 0.3 ms/year in SCD cases and controls, respectively (P = 0.147). Changes in PR and QTc interval were not significantly different over time.
Bundle branch block and intraventricular conduction delay
In the majority of patients with sequential ECGs, increase in QRS duration was due to non-specific intraventricular conduction delay. Twenty-five out of 120 SCD cases (21%) and 22 out of the 242 controls (9%) had complete RBBB. Complete LBBB was rare, only 3 out of the 120 SCD cases (3%) and 6 out of the 242 controls (2%) had complete LBBB. During follow-up, new onset of RBBB occurred in 10 patients (5 SCD cases and 5 controls). New onset of LBBB occurred only in two control patients.
Cardiac surgery had been performed in 44 patients (9 SCD cases and 35 controls) during the study period of 7 ± 5 years. The majority of these patients (84%) underwent pulmonary or aortic valve replacement. A new and persistent RBBB after cardiac surgery occurred only in two control patients [one patient with a pulmonary valve replacement (PVR) with homograft and aortic valve replacement and one patient with isolated PVR with homograft].
Echocardiographic parameters
Overall, 81 of the 131 SCD cases (62%) and 201 out of the 260 controls (77%) had sequential echocardiograms. At last follow-up, 34% of SCD cases had a moderately to severely impaired SVF vs. 14% of controls (P < 0.001). Subpulmonary ventricular function (SPVF) was moderately to severely impaired in 19 vs. 8% of SCDs vs. controls (P < 0.001), respectively. The association of annual changes in ventricular function with the risk of SCD is summarized in Table 2. The risk of SCD was higher if impairment of SVF progressed by at least one point per year. Importantly, the risk of SCD increased if SVF changed from normal or mild dysfunction to moderate or severe dysfunction (OR 2.42, 95% CI 1.21–4.84, P = 0.012). Change from any baseline SVF (normal, mild, or moderately impaired) to severe systemic dysfunction was associated with the highest risk of SCD (9.9% of SCDs and 0.5% of controls, OR 16.9, 95% CI 1.8–120.1, P = 0.008). Change from normal to mild systemic dysfunction (OR 0.45, 95% CI 0.20–0.99, P = 0.050) was not associated with increased risk of SCD. The same results were observed for changes in SPVF: deterioration of any baseline SPVF (normal, mild, or moderate) to severe subpulmonary dysfunction was associated with increased risk of SCD (4.9% of SCDs and 0.5% of controls, OR 8.2, 95% CI 0.9–73.7, P = 0.027).
Association of changes in ventricular function per year (p/y) with the risk of SCD
| SCDs (n = 81) | Controls (n = 201) | OR | 95% CI | P | |
|---|---|---|---|---|---|
| SVF, n (%) | |||||
| Impairment of SVF p/y | 22 (27) | 45 (22) | 1.3 | 0.78–2.19 | 0.035 |
| Impairment of SVF ≥1 point p/y | 11 (14) | 10 (5) | 3.0 | 1.30–6.76 | 0.010 |
| Unchanged SVF | 45 (56) | 126 (63) | 0.9 | 0.54–1.36 | 0.054 |
| Improvement of SVF p/y | 14 (17) | 30 (15) | 0.8 | 0.48–1.79 | 0.826 |
| Improvement of SVF ≥1 point p/y | 4 (5) | 12 (6) | 0.7 | 0.22–2.22 | 0.543 |
| SPVF, n (%) | |||||
| Impairment of SPVF p/y | 25 (31) | 38 (19) | 1.7 | 0.95–3.07 | 0.075 |
| Impairment of SPVF ≥1 point p/y | 10 (12) | 7 (3) | 3.6 | 1.30–9.68 | 0.013 |
| Unchanged SPVF | 51 (63) | 123 (61) | 0.6 | 0.50–1.45 | 0.504 |
| Improvement of SPVF p/y | 8 (10) | 30 (15) | 0.5 | 0.25–1.29 | 0.176 |
| Improvement of SPVF ≥1 point p/y | 3 (4) | 7 (3) | 0.9 | 0.25–3.86 | 0.970 |
| SCDs (n = 81) | Controls (n = 201) | OR | 95% CI | P | |
|---|---|---|---|---|---|
| SVF, n (%) | |||||
| Impairment of SVF p/y | 22 (27) | 45 (22) | 1.3 | 0.78–2.19 | 0.035 |
| Impairment of SVF ≥1 point p/y | 11 (14) | 10 (5) | 3.0 | 1.30–6.76 | 0.010 |
| Unchanged SVF | 45 (56) | 126 (63) | 0.9 | 0.54–1.36 | 0.054 |
| Improvement of SVF p/y | 14 (17) | 30 (15) | 0.8 | 0.48–1.79 | 0.826 |
| Improvement of SVF ≥1 point p/y | 4 (5) | 12 (6) | 0.7 | 0.22–2.22 | 0.543 |
| SPVF, n (%) | |||||
| Impairment of SPVF p/y | 25 (31) | 38 (19) | 1.7 | 0.95–3.07 | 0.075 |
| Impairment of SPVF ≥1 point p/y | 10 (12) | 7 (3) | 3.6 | 1.30–9.68 | 0.013 |
| Unchanged SPVF | 51 (63) | 123 (61) | 0.6 | 0.50–1.45 | 0.504 |
| Improvement of SPVF p/y | 8 (10) | 30 (15) | 0.5 | 0.25–1.29 | 0.176 |
| Improvement of SPVF ≥1 point p/y | 3 (4) | 7 (3) | 0.9 | 0.25–3.86 | 0.970 |
OR, odds ratio; CI, confidence interval; SVF, systemic ventricular function; SPVF, subpulmonary ventricular function.
Association of changes in ventricular function per year (p/y) with the risk of SCD
| SCDs (n = 81) | Controls (n = 201) | OR | 95% CI | P | |
|---|---|---|---|---|---|
| SVF, n (%) | |||||
| Impairment of SVF p/y | 22 (27) | 45 (22) | 1.3 | 0.78–2.19 | 0.035 |
| Impairment of SVF ≥1 point p/y | 11 (14) | 10 (5) | 3.0 | 1.30–6.76 | 0.010 |
| Unchanged SVF | 45 (56) | 126 (63) | 0.9 | 0.54–1.36 | 0.054 |
| Improvement of SVF p/y | 14 (17) | 30 (15) | 0.8 | 0.48–1.79 | 0.826 |
| Improvement of SVF ≥1 point p/y | 4 (5) | 12 (6) | 0.7 | 0.22–2.22 | 0.543 |
| SPVF, n (%) | |||||
| Impairment of SPVF p/y | 25 (31) | 38 (19) | 1.7 | 0.95–3.07 | 0.075 |
| Impairment of SPVF ≥1 point p/y | 10 (12) | 7 (3) | 3.6 | 1.30–9.68 | 0.013 |
| Unchanged SPVF | 51 (63) | 123 (61) | 0.6 | 0.50–1.45 | 0.504 |
| Improvement of SPVF p/y | 8 (10) | 30 (15) | 0.5 | 0.25–1.29 | 0.176 |
| Improvement of SPVF ≥1 point p/y | 3 (4) | 7 (3) | 0.9 | 0.25–3.86 | 0.970 |
| SCDs (n = 81) | Controls (n = 201) | OR | 95% CI | P | |
|---|---|---|---|---|---|
| SVF, n (%) | |||||
| Impairment of SVF p/y | 22 (27) | 45 (22) | 1.3 | 0.78–2.19 | 0.035 |
| Impairment of SVF ≥1 point p/y | 11 (14) | 10 (5) | 3.0 | 1.30–6.76 | 0.010 |
| Unchanged SVF | 45 (56) | 126 (63) | 0.9 | 0.54–1.36 | 0.054 |
| Improvement of SVF p/y | 14 (17) | 30 (15) | 0.8 | 0.48–1.79 | 0.826 |
| Improvement of SVF ≥1 point p/y | 4 (5) | 12 (6) | 0.7 | 0.22–2.22 | 0.543 |
| SPVF, n (%) | |||||
| Impairment of SPVF p/y | 25 (31) | 38 (19) | 1.7 | 0.95–3.07 | 0.075 |
| Impairment of SPVF ≥1 point p/y | 10 (12) | 7 (3) | 3.6 | 1.30–9.68 | 0.013 |
| Unchanged SPVF | 51 (63) | 123 (61) | 0.6 | 0.50–1.45 | 0.504 |
| Improvement of SPVF p/y | 8 (10) | 30 (15) | 0.5 | 0.25–1.29 | 0.176 |
| Improvement of SPVF ≥1 point p/y | 3 (4) | 7 (3) | 0.9 | 0.25–3.86 | 0.970 |
OR, odds ratio; CI, confidence interval; SVF, systemic ventricular function; SPVF, subpulmonary ventricular function.
Impairment of SPVF by at least one point per year was associated with a nearly four-fold increased risk of SCD. Deterioration of SPVF over time was consistently more frequent in SCD cases than in controls irrespective of the underlying cardiac lesions. Impairment of SPVF was seen in all SCD cases (n = 15) with cyanotic non-Eisenmenger syndrome and 4 out of 29 matched controls (14%) with same diagnosis (P = 0.011). In other cardiac lesions, impairment of SPVF occurred in ∼30% of the SCD cases, which was not significantly different compared with controls.
There was no significant correlation between changes in QRS duration and changes in SPVF or SVF.
Cardiac surgery had been performed in 32 patients (10 SCDs and 22 controls) with sequential echocardiograms. Surgery led to improvement of SVF in three patients (one SCD, two controls) and improvement in SPVF in six patients (two SCDs, four controls). Deterioration of SVF occurred in 15 patients after cardiac surgery (6 SCD cases, 9 controls), and deterioration of SPVF occurred in 10 patients (3 SCDs and 7 controls). Exclusion of patients who had undergone surgery did not alter the risk of SCD.
Discussion
In an overall CHD population, we observed an increase in QRS duration and progression of ventricular dysfunction over time. Both the rate and the magnitude of these changes correlated with an incremental risk of SCD. Shortly before SCD, QRS duration and QT dispersion crossed the upper limit of normal values (QRS > 120 ms, QT dispersion > 70 ms). Deterioration of SPVF over time occurred significantly more often in SCD cases with cyanotic non-Eisenmenger syndrome. Impairment of SVF, although not significant, was more frequently seen in SCD cases with Fontan circulation, the Eisenmenger syndrome, left-sided lesions, closed septal defects, and cyanotic non-Eisenmenger patients.
The relation between QRS duration, increased risk of SCD, and life-threatening arrhythmias has been reported earlier. This relates to a point estimate in time. In patients with prior myocardial infarction or medically refractory heart failure, QRS duration is a significant and independent predictor of cardiovascular mortality.17 Prolongation of QRS duration over time has been shown to be an even better predictor of survival in elderly patients with heart failure.12 However, little is known about prognostic meaning of changes in electrocardiographic parameters over time in the population of adults with CHD. Progressive QRS prolongation was observed in a cohort of 35 tetralogy of Fallot (TOF) patients who received transannular patches but also in those who received no patch or a pulmonary homograft during long-term follow-up.18 Another study found that a QRS duration of >180 ms, either pre- or postoperatively, or the lack of a reduction in QRS duration after PVR in TOF was associated with a worse outcome.19,20 These results are consistent with our findings.
In this study, prolongation of QRS duration was predominantly due to non-specific intraventricular conduction delay. New onset of complete RBBB or LBBB was rare and caused by surgical injury in only minority of these patients. These findings suggest that progressive broadening of the QRS complex might be caused by worsening of ventricular function. This was also the case in control patients. However, increase in QRS duration over time was less progressive in these patients. Increase in QRS duration occurred significantly more often in SCD cases compared with controls and especially seen in patients with the Eisenmenger syndrome and closed septal defects. Differences in other diagnostic subgroups were not significant, possibly due to the small number of patients. These findings are in line with a previous study on risk factors for arrhythmia and SCD late after repair of TOF. In that study, a significantly greater QRS rate of change was seen between 1985 and the year before clinical events in the ventricular tachycardia and SCD groups.6
Left ventricular dysfunction is one of the most powerful predictors of mortality in various acquired heart disease.21 To the best of our knowledge, few data exist on how ventricular function changes over time or how changes from a given baseline may be used to assess the subsequent clinical course or predict survival. In congestive heart failure, there are conflicting results on changes in left ventricular EF over time and its effect on outcome.22,23
Similarly, in adults with CHD, a single measurement of systemic or subpulmonary ventricular dysfunction has been associated with clinical deterioration and increased mortality in various studies.5–7,24 In this study, we found that aggravation in impairment of systemic or SPVF over time was associated with an increased risk of SCD. Deterioration of systemic or SPVF from normal, mild, or moderate dysfunction to severe dysfunction over time was associated with the highest risk of SCD and the only clinically relevant finding. However, since the only systemic ventricular change associated with a significantly increased risk of SCD was a change from any class to severe, the opportunity to intervene in a meaningful capacity clinically at this stage may be limited. Ventricular function might have been influenced during follow-up as a result of therapeutic interventions. The magnitude of this effect is unknown, but a limited number of patients had undergone cardiac surgery and the incidence of surgery was not different between SCD cases and controls. This finding indicates that therapeutic intervention of SVF is not a causal determinant of mortality due to SCD.5
Clinical implications
Apart from the absolute values of the parameters studied, the dynamic changes appear equally interesting in the assessment of SCD risk in patients with CHD. Specific attention and documentation of absolute changes in electrocardiographic and echocardiographic parameters may be considered in everyday practice. Further studies are required to investigate whether rapid increase in QRS duration and QT dispersion and ventricular deterioration justify implantation of an implantable cardioverter defibrillator.
Study limitations
This study has limitations inherent to any retrospective study. Data collection was limited to data available in medical records. We were not able to retrieve ECGs and echocardiograms in all patients during all time frames. Also, we have not assessed inter- and intra-observer variability for the ECG and echocardiographic measurements. Assessment of ventricular function may be challenging in CHD due to abnormal ventricular geometry. Cardiac magnetic resonance imaging (cMRI) is more accurate in measuring EF in some cardiac lesions (e.g. systemic right ventricle). Unfortunately, in this study, quantification of ventricular EF by cMRI was frequently lacking, and therefore qualitative data on ventricular function from echocardiograms were used. The use of antiarrhythmic drugs or angiotensin converting enzyme inhibitors during the study period may have influenced QRS duration and ventricular function. In the databases used, comprehensive data on medication use at different time frames were frequently incomplete.
Conclusion
In the population of adults with CHD, QRS duration and ventricular dysfunction change over time in many patients. The rate—as well as magnitude—of increase in the QRS duration and the rate at which both systemic and subpulmonary ventricular dysfunction deteriorate serve to identify those at increased risk of SCD, meriting heightened surveillance.
Supplementary material
Supplementary material is available at Europace online.
Funding
The work described in this study was carried out in the context of the Parelsnoer Institute (PSI). The Parelsnoer Institute is part of and funded by the Dutch Federation of University Medical Centers. J.R.d.G. is supported by VIDI grant from NWO/ZonMW 016146310.
Conflict of interest: none declared.
