Abstract
Aims
To evaluate the prognostic significance of novel P-wave morphology descriptors in general population.
Methods and results
Novel P-wave morphology variables were analyzed from orthogonal X-, Y-, Z-leads of the digitized electrocardiogram using a custom-made software in 6906 middle-aged subjects of the Mini-Finland Health Survey. A total of 3747 (54.3%) participants died during the follow-up period of 24.3 ± 10.4 years; 379 (5.5%) of the study population succumbed to sudden cardiac death (SCD), 928 (13.4%) to non-SCD (NSCD) and 2440 (35.3%) patients to non-cardiac death (NCD). In univariate comparisons, most of the studied P-wave morphology parameters had a significant association with all modes of death (P from <0.05 to <0.001). After relevant adjustments in the Cox multivariate hazards model, P-wave morphology dispersion (PMD) still tended to predict SCD [hazard ratio (HR): 1.006, 95% confidence interval (CI): 1.000–1.012, P = 0.05) but not NSCD (HR: 0.999, 95% CI: 0.995–1.003, P = 0.68) or NCD (HR: 0.999, 95% CI: 0.997–1.001, P = 0.44). The P-wave maximum amplitude in the lead Z (P-MaxAmp-Z) predicted SCD even after multivariate adjustments (HR: 1.010, 95% CI: 1.005–1.015, P = 0.0002) but also NSCD (HR: 1.005, 95% CI: 1.002–1.009, P = 0.0005) and NCD (HR: 1.002, 95% CI: 1.000–1.005, P = 0.03).
Conclusion
Abnormalities of P-wave morphology are associated with the risk of all modes of death in general population. After relevant adjustments, PMD was still closely associated with the risk of SCD but not with NSCD or NCD. P-MaxAmp-Z predicted SCD even after adjustments, however, it also retained its association with NSCD and NCD.
Keywords: Electrocardiogram, P-wave morphology, Sudden cardiac death, General population
What’s new?
P-wave morphology parameters yield long-term prognostic information in general population
New novel P-wave morphology parameters have an association with the risk of sudden cardiac death
Introduction
Abnormalities in the P-wave are a relatively common electrocardiographic finding in middle-aged subjects.1 Abnormalities of P-wave have been observed to be associated with factors, such as elevated left atrial pressure2 and left ventricular interstitial fibrosis,3 suggesting that P-wave changes may contain prognostic information. The interest has often focused on conventional measurements of P-wave. Conventional P-wave measurements, including P-wave duration, P-terminal force, and P-wave axis, have been shown to predict mortality in general population.1,4,5 Prolongation of P-wave and P-terminal force have also been attributed with the risk of sudden cardiac death (SCD) in general population,6,7 and P-terminal force with the risk of development of atrial fibrillation.1 Orthogonal P-wave morphology has recently also shown to be associated with the risk of hospitalization due to atrial fibrillation in general population.8 P-terminal force has been associated with risk of cardiac death after acute myocardial infarction9 and novel, sophisticated P-wave morphology parameters have been found to predict independently SCD in patients with coronary artery disease.10 However, the prognostic significance of novel P-wave morphology descriptors in general population is not well established. Therefore, we evaluated the value of novel P-wave morphology variables in predicting different modes of death in 6906 middle-aged subjects of the Mini-Finland Health Survey who were followed up 24.3 ± 10.4 years.
Methods
The general population sample consisted of subjects who participated in The Mini-Finland Health Survey, (1978–1980), which was a part of the Social Insurance Institution’s Finnish Mobile Clinic Survey.11 The study protocol details and methods are described previously.12 The aim of the study was to create a versatile picture of the health of Finnish adults, survey the need for health care and develop the population’s health measurement and monitoring methods. The Mini-Finland Health Survey was based on a representative sample of the Finnish population and included 8000 subjects aged ≥30 years. At the beginning of the survey, 7703 subjects had a home visit interview made by a nurse and 7217 participated in the basic health examination. Before the basic health examination, each participant was interviewed about their health status, known diseases, smoking and alcohol habits, and medication. The basic health examination included blood pressure measurement, weight and height measurements, body mass index, total serum cholesterol, high-density lipoprotein cholesterol and triglyceride levels and electrocardiogram measurements. The resting electrocardiogram strips were compiled to the Biological sample bank. The follow-up length was up to 33 years (1978–2011) and the information from nationwide health registers was linked to the survey.12 During the follow-up, a total of 3967 deaths occurred. All subjects participated in the survey voluntarily and were fully informed about the study. The Social Insurance Institution, the Finnish Institute for Health and Welfare, and Statistics Finland had approved the record linkage of national health registers.
The endpoints of the present study were SCD, non-SCD (NSCD), and non-cardiac death (NCD). Sudden cardiac death was defined as witnessed death within 1 hour after the symptoms began or death within 24 h of last seen alive. Sudden cardiac death definitions were made by using the CAST-criteria.13 Non-SCD was defined as cardiac death that did not fulfil the criteria of SCD.
In 1978–1980, the resting electrocardiogram was registered with a device which recorded both standard 12-lead electrocardiogram recorded in paper format with a paper speed of 50 mm/s and calibration of 10 mm/mV and orthogonal Frank leads (X, Y, and Z) recorded for ∼10 s.13 After 254 electrocardiograms were missing or excluded, the total digitized number of electrocardiograms were 6963.
Later the paper sheets were scanned into single Multi-Page TIFF files, and the files were digitized with a proprietary application, the ECG Trace Tool.14 The subjects with no X-, Y-, Z-data and the subjects, whose P-wave analysis was not possible mostly because of atrial fibrillation, atrial flutter, non-sinus rhythm or technical problems, were excluded, leaving 6906 subjects for the present data analysis.
P-wave morphology parameters were calculated from the orthogonal Frank (X-, Y-, and Z-) leads using custom-made electrocardiogram analysis software. The software enables semi-automated analysis of digitized electrocardiogram recordings. It creates low-noise representative median beats for each of the twelve leads, trimmed mean beats for the X-, Y-, Z-leads and automatically locates the P-wave and its boundaries, which are marked by adjustable calipers.14 Premature beats and complexes with excessive artefacts were removed for representative beat formation. After visual verification and manual adjustment of calipers (if required), electrocardiogram analysis software computed several conventional and advanced electrocardiographic and vectorcardiographic parameters and saved the results in spreadsheet format for subsequent statistical analyses.14
Several P-wave loop parameters were calculated automatically and blindly from the averaged orthogonal Frank leads (X, Y, Z). P-wave duration (P-dur) was measured from the beginning of the P-wave to the end of the P-wave. P-wave heterogeneity (PWH) was calculated using second central moment analysis.15 Second central moment is a concept from Newtonian mechanics that refers to a measure of splay of waveforms around the first moment, which in this case is the average waveform of P- wave calculated from leads X, Y, and Z (Figure 1). High values of PWH indicate heterogeneity between the P-wave morphologies across the leads, whereas low values of PWH indicate similar morphologies across the leads. P-wave morphology dispersion (PMD degrees) expresses the variation of P-wave morphology between individual leads in degrees, by calculating the average angle between all reconstruction vector pairs in the P-wave loop as previously done in T-wave analysis16 (Figure 1). Higher values of PMD describe abnormal variation in atrial depolarization. In the principal component analysis, the first dimension of the P-wave loop (P-PCA1) represents the length of the loop, the second dimension (P-PCA2) represents the height of the loop in the preferential plane and the third dimension of the P-wave loop (P-PCA3) represents the three-dimensional deviation from the 2-dimensional preferential plane. Other features were the maximum positive amplitude of the P-wave in lead X (P-MaxAmp-X), in lead Y (P-MaxAmp-Y) and in lead Z (P-MaxAmp-Z).
Figure 1.
Panel A contains simultaneous superimposed excerpts from leads X, Y, and Z, illustrating P-wave heterogeneity (PWH) as interlead splay around mean atrial depolarization morphology. Maximum P-wave amplitudes are also calculated from this view for each lead. Panel B illustrates P-wave morphology dispersion (PMD), which is obtained by calculating the average angle between all possible reconstruction vector pairs of limb leads I-II and chest leads V2-V6 in three-dimensional space.
The Mann–Whitney U-test was used to compare the continuous non-normally distributed variables between the groups, the t-test was used for normally distributed variables, respectively. The Fisher’s test was used to compare the categorical variables’ prevalence. The baseline characteristics, which differed at the level P < 0.05 between study participants with certain mode of death and those without such an event, were tested in the multivariate Cox regression analysis using the stepwise backward method. All the P-wave parameters were tested in the corresponding Cox clinical hazard model related to each death class one at a time. The follow-up times were applied for each participant. The C-index was calculated to evaluate the accuracy of the clinical model with PMD and P-MaxAmp-Z in predicting the occurrence of SCD. The association of clinical risk markers with the P-wave morphology parameters was evaluated by determining the Pearsons’ correlation coefficients and by assessing the statistical significance of the difference of the values of P-morphology parameters between the study participants with or without a categorized clinical characteristic. The Kaplan–Meier curves were determined to show the cumulative proportional probabilities of SCD-free survival, NSCD-free survival, and NCD-free survival in the study participants according to the dichotomized values of PMD and P-MaxAmp-Z. The cut-off points were optimized from the receiver operating characteristics (ROC) curves. The statistical significance of the separation of the curves was assessed using the log-rank test. A P-value <0.05 was considered as statistically significant. IBM SPSS version 25 (IBM, Armonk, NY, USA) was used for statistical analyses.
Results
The P-wave morphology could be analyzed in 6906 subjects. A total of 3747 (54.3%) of the present study participants died, of whom 379 (5.5%) succumbed to SCD, 928 (13.4%) to NSCD, and 2440 (35.3%) to NCD during the follow-up period of 24.3 ± 10.4 years (Table 1). The causes of NCD included malignant diseases (38.2%), cerebral infarction/haemorrhage (16.9%), end-stage dementia (10.1%), infections (8.6%), trauma/accident (5.0%), chronic pulmonary disease (4.3%), severe neurologic disease (2.5%), abdominal vascular disease (2.2%), chronic liver disease (1.4%), end-stage renal disease (1.3%), intoxication (1.2%), pulmonary embolism (1.0%), and peripheral artery disease (0.9%).
Table 1.
Baseline characteristics of the study subjects
| Variable | Alive n = 3159 |
SCD n = 379 |
NSCD n = 928 |
NCD n = 2440 |
|---|---|---|---|---|
| Age (years) | 41.2 ± 8.0 | 57.1 ± 12.3‡ | 62.7 ± 11.1‡ | 58.9 ± 12.5‡ |
| Male gender (%) | 41.7 | 66.8‡ | 48.0 | 46.9 |
| BMI (kg/m2) | 25.0 ± 3.7 | 26.7 ± 4.2‡ | 26.9 ± 4.2‡ | 26.5 ± 4.3‡ |
| HRa (bpm) | 65.8 ± 12.0 | 70.8 ± 15.4‡ | 69.9 ± 13.8‡ | 69.3 ± 14.0‡ |
| SBP (mmHg) | 132.6 ± 16.4 | 153.2 ± 23.6‡ | 157.5 ± 24.6‡ | 150.1 ± 23.5‡ |
| DBP (mmHg) | 84.4 ± 10.6 | 90.1 ± 12.1‡ | 89.5 ± 11.7‡ | 88.5 ± 11.7‡ |
| DM, (%) | 0.7 | 8.2* | 14.1‡ | 7.7‡ |
| Smoker (%) | 20.8 | 39.6‡ | 20.7* | 26.3‡ |
| Alcohol (g/wk) | 46.0 ± 99.0 | 67.6 ± 133.7* | 33.7 ± 97.3‡ | 45.6 ± 116.3‡ |
| fP-gluc (mmol/l) | 5.12 ± 0.63 | 5.68 ± 1.94† | 5.92 ± 2.20‡ | 5.59 ± 1.67‡ |
| fS-chol (mmol/l) | 6.67 ± 1.27 | 7.47 ± 1.38‡ | 7.34 ± 1.39‡ | 7.12 ± 1.41‡ |
| fS-HDL-chol (mmol/l) | 1.75 ± 0.39 | 1.61 ± 0.39‡ | 1.61 ± 0.42‡ | 1.67 ± 0.43‡ |
| fP-crea (μmol/l) | 70.9 ± 11.2 | 75.0 ± 12.5† | 74.4 ± 15.1‡ | 74.0 ± 19.7‡ |
| fS-trigly (mmol/l) | 1.26 ± 0.71 | 1.78 ± 0.98‡ | 1.90 ± 1.23‡ | 1.71 ± 1.32‡ |
| LBBB (%) | 0.1 | 1.3 | 1.7‡ | 0.7 |
| RBBB (%) | 0.2 | 1.8* | 2.0‡ | 1.1 |
| MI (%) | 0.2 | 12.1‡ | 11.5‡ | 4.5 |
| AP (%) | 2.0 | 21.9‡ | 25.2‡ | 11.7‡ |
| CHD (%) | 0.2 | 0 | 0.1 | 0 |
| VHD (%) | 0.2 | 1.1 | 2.2‡ | 0.8 |
| Hypertension (%) | 11.4 | 33.0‡ | 36.1‡ | 28.2‡ |
| HF (%) | 0.6 | 14.2‡ | 22.2‡ | 12.4‡ |
| HHD (%) | 0.4 | 4.5 | 7.1‡ | 4.8‡ |
| PHD (%) | 0 | 0 | 0.2 | 0.8‡ |
| ASO (%) | 0.1 | 5.5‡ | 3.2‡ | 1.9* |
| CVD (%) | 0.3 | 1.8 | 3.4‡ | 2.3‡ |
The study participants who experienced certain mode of death were compared with those without such an event. The values are mean ± SD or percentages.
Age, age at the time of the clinical examination; alcohol, alcohol consumption grams per week; AP, suspected/confirmed angina pectoris; ASO, suspected/confirmed arteriosclerosis obliterans of lower limbs; BMI, body mass index; CHD, suspected/confirmed congenital heart defect; CVD, suspected/confirmed cerebral vascular disease; DBP, diastolic blood pressure; DM, diabetes mellitus; fS-chol, fasting serum cholesterol, fP-crea, fasting plasma creatinine; fP-gluc, fasting plasma glucose; fS-HDL-chol, fasting serum high-density lipoprotein cholesterol; fS-trigly, fasting serum triglycerides; HF, suspected/confirmed heart failure; HHD, suspected/confirmed hypertensive heart disease; HRa, heart rate, LBBB, left bundle branch block; MI, suspected/confirmed myocardial infarction, NCD, non-cardiac death; NSCD, non-sudden cardiac death, PHD, suspected/confirmed pulmonary heart disease, RBBB, right bundle branch block; SBP, systolic blood pressure, SCD, sudden cardiac death, VHD, suspected/confirmed valvular heart disease.
P< 0.05.
P< 0.01.
P< 0.001.
The baseline characteristics of the study population are shown in Table 1. Study participants who experienced SCD were older, more commonly males and smokers, had higher body mass index, heart rate, systolic and diastolic blood pressure, fasting plasma glucose and creatinine, fasting serum cholesterol and triglycerides, and lower fasting serum high-density lipoprotein cholesterol, consumed more alcohol, had more commonly diabetes, right bundle branch block, suspected or confirmed myocardial infarction, angina pectoris, hypertension, heart failure and suspected/confirmed arteriosclerosis obliterans of lower limbs, compared with those who did not experience SCD during the follow-up. The clinical characteristics associated with NSCD and NCD are shown in Table 1.
The P-wave morphology parameters are shown in Table 2. The study participants who experienced either SCD, NSCD or NCD had longer P-dur, higher P-MaxAmp-X and P-MaxAmp-Z, higher values of P-PCA2, P-PCA3, and PMD, and lower value of P-PCA1. The study participants who succumbed to SCD or NCD had higher values of PWH.
Table 2.
P-wave morphology parameters in study patients at baseline
| Variable | Alive n = 3159 | SCD n = 379 | NSCD n = 928 | NCD n = 2440 |
|---|---|---|---|---|
| P-dur (ms) | 119.6 ± 13.6 | 125.9 ± 14.4‡ | 126.5 ± 14.8‡ | 124.8 ± 15.0‡ |
| P-MaxAmp-X (µV) | 57.8 ± 21.7 | 63.7 ± 22.5* | 63.4 ± 23.4† | 63.5 ± 26.4‡ |
| P-MaxAmp-Y (µV) | 86.1 ± 40.2 | 89.5 ± 39.9 | 84.1 ± 36.8 | 86.6 ± 39.1 |
| P-MaxAmp-Z (µV) | 20.2 ± 14.1 | 35.2 ± 25.0‡ | 35.0 ± 22.8‡ | 30.3 ± 20.4‡ |
| P-PCA1 | 316.5 ± 142.7 | 255.3 ± 116.7‡ | 246.8 ± 114.5‡ | 264.5 ± 127.9‡ |
| P-PCA2 | 34.7 ± 14.5 | 43.0 ± 16.5‡ | 44.6 ± 17.4‡ | 42.0 ± 17.2‡ |
| P-PCA3 | 12.8 ± 6.13 | 15.1 ± 7.29† | 15.8 ± 7.55‡ | 15.0 ± 7.09‡ |
| PMD (degrees) | 40.8 ± 17.8 | 47.0 ± 18.8‡ | 46.2 ± 18.8‡ | 44.9 ± 17.9‡ |
| PWH | 23.6 ± 7.52 | 25.1 ± 8.05† | 24.4 ± 7.71 | 24.4 ± 8.06† |
The study participants who experienced certain mode of death were compared with those without such an event. The values are mean ± SD.
NCD, non-cardiac death; NSCD, non-sudden cardiac death; P-dur, P-wave duration; P-MaxAmp-X, the maximum positive amplitude of the P-wave in lead X; P-MaxAmp-Y, the maximum positive amplitude of the P-wave in lead Y; P-MaxAmp-Z, the maximum positive amplitude of the P-wave in lead Z; PMD, P-wave morphology dispersion; P-PCA1, the first dimension of the principal component analysis of the P-wave loop; P-PCA2, the second dimension of the principal component analysis of the P-wave loop; P-PCA3, the third dimension of the principal component analysis of the P-wave loop; PWH, P-wave heterogeneity; SCD, sudden cardiac death. Please see the Methods section for details.
P< 0.05.
P< 0.01.
P< 0.001.
Only the clinical variables that differed significantly in the comparison between study participants with certain mode of death and those without such an event were included in the Cox multivariate analysis (Table 1). The variables that remained statistically significant in the multivariate analysis were included in the clinical model for each mode of death. All the P-wave morphology variables were tested one at a time in the corresponding clinical model.
When the clinical variables that differed significantly between study participants who experienced SCD and those without such an event (Table 1) were tested in the Cox multivariate model, age, male gender, heart rate, systolic blood pressure, smoking, fasting plasma glucose, fasting serum cholesterol, a history of myocardial infarction, a history of angina pectoris, and arteriosclerosis obliterans of lower limbs remained significant predictors of SCD after adjustments (Table 3). The clinical variables that remained significant in the clinical model for NSCD and NCD after adjustments are shown in Table 3. When the P-wave morphology parameters were tested one at a time in the clinical models of different modes of death, PMD was still closely associated with SCD, but not with NSCD or NCD. P-MaxAmp-Z predicted SCD even after adjustments, however, it also retained its association with NSCD and NCD. In the multivariate model, P-dur, P-MaxAmp-X and PWH retained significant association with NCD, and P-PCA3 with NSCD and NCD (Table 4).
Table 3.
Clinical characteristics predicting sudden cardiac death, non-sudden cardiac death, and non-cardiac death in the cox hazards model
| Variable | SCD n = 379 | NSCD n = 928 | NCD n = 2440 | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P-value | HR | 95% CI | P-value | HR | 95% CI | P-value | ||
| Age | uv | 1.075 | 1.066–1.084 | <0.000001 | 1.136 | 1.129–1.144 | <0.000001 | 1.105 | 1.101–1.109 | <0.000001 |
| mv | 1.071 | 1.060–1.083 | <0.000001 | 1.128 | 1.119–1.137 | <0.000001 | 1.111 | 1.107–1.116 | <0.000001 | |
| Male | uv | 2.617 | 2.113–3.240 | <0.000001 | ||||||
| mv | 3.228 | 2.547–4.091 | <0.000001 | |||||||
| SBP | uv | 1.028 | 1.024–1.031 | <0.000001 | 1.035 | 1.032–1.037 | <0.000001 | |||
| mv | 1.013 | 1.008–1.018 | <0.000001 | 1.008 | 1.005–1.012 | 0.000006 | ||||
| DBP | uv | 1.028 | 1.022–1.034 | <0.000001 | 1.021 | 1.017–1.024 | <0.000001 | |||
| mv | 1.009 | 1.002–1.016 | 0.01 | 1.009 | 1.005–1.013 | <0.000001 | ||||
| HRa | uv | 1.022 | 1.015–1.029 | <0.000001 | 1.015 | 1.012–1.018 | <0.000001 | |||
| mv | 1.017 | 1.010–1.024 | 0.000004 | 1.004 | 1.001–1.007 | 0.007 | ||||
| DM | uv | 6.780 | 5.614–8.188 | <0.000001 | 3.660 | 3.150–4.254 | <0.000001 | |||
| mv | 1.818 | 1.423–2.322 | 0.000002 | 1.409 | 1.163–1.707 | <0.000001 | ||||
| Smoker | uv | 2.239 | 1.822–2.751 | <0.000001 | 0.898 | 0.766–1.053 | 0.18 | 1.230 | 1.124–1.346 | 0.000007 |
| mv | 2.787 | 2.221–3.497 | <0.000001 | 2.068 | 1.740–2.458 | <0.000001 | 2.076 | 1.883–2.289 | <0.000001 | |
| Alcohol | uv | 0.998 | 0.997–0.999 | 0.0002 | 0.9999 | 0.999–1.000 | 0.55 | |||
| mv | 1.001 | 1.000–1.002 | 0.01 | 1.001 | 1.0008–1.001 | <0.000001 | ||||
| fP-gluc | uv | 1.146 | 1.105–1.188 | <0.000001 | 1.180 | 1.159–1.202 | <0.000001 | 1.144 | 1.127–1.162 | <0.000001 |
| mv | 1.091 | 1.029–1.156 | 0.003 | 1.105 | 1.065–1.146 | <0.000001 | 1.046 | 1.011–1.081 | 0.009 | |
| fS-chol | uv | 1.347 | 1.261–1.439 | <0.000001 | 1.280 | 1.226–1.337 | <0.000001 | 1.159 | 1.127–1.193 | <0.000001 |
| mv | 1.210 | 1.126–1.299 | <0.000001 | 1.073 | 1.022–1.127 | 0.005 | 0.945 | 0.915–0.976 | 0.0006 | |
| fS-HDL-chol | uv | 0.467 | 0.392–0.557 | <0.000001 | ||||||
| mv | 0.663 | 0.558–0.789 | 0.000003 | |||||||
| fP-crea | uv | 1.010 | 1.008–1.012 | <0.000001 | 1.010 | 1.008–1.012 | <0.000001 | |||
| mv | 1.007 | 1.004–1.010 | 0.00004 | 1.007 | 1.005–1.010 | <0.000001 | ||||
| fS-trigly | uv | 1.116 | 1.102–1.130 | <0.000001 | ||||||
| mv | 1.048 | 1.017–1.081 | 0.002 | |||||||
| LBBB | uv | 7.283 | 4.437–11.955 | <0.000001 | ||||||
| mv | 2.245 | 1.359–3.710 | 0.002 | |||||||
| MI | uv | 7.837 | 5.725–10.728 | <0.000001 | 7.949 | 6.473–9.763 | <0.000001 | |||
| mv | 2.370 | 1.640–3.425 | 0.000004 | 2.160 | 1.720–2.711 | <0.000001 | ||||
| AP | uv | 4.851 | 3.788–6.212 | <0.000001 | 6.206 | 5.336–7.218 | <0.000001 | |||
| mv | 1.855 | 1.387–2.480 | 0.00003 | 1.741 | 1.472–2.059 | <0.000001 | ||||
| VHD | uv | 5.553 | 3.563–8.655 | <0.000001 | ||||||
| mv | 2.975 | 1.898–4.663 | 0.000002 | |||||||
| PHD | uv | 9.435 | 5.998–14.839 | <0.000001 | ||||||
| mv | 2.074 | 1.313–3.274 | 0.002 | |||||||
| ASO | uv | 10.048 | 6.435–15.687 | <0.000001 | 3.953 | 2.958–5.285 | <0.000001 | |||
| mv | 1.666 | 1.034–2.685 | 0.04 | 1.412 | 1.049–1.900 | 0.02 | ||||
| CVD | uv | 5.366 | 3.766–7.648 | <0.000001 | 3.724 | 2.848–4.869 | <0.000001 | |||
| mv | 1.601 | 1.103–2.323 | 0.01 | 1.581 | 1.205–2.076 | 0.001 | ||||
Age, age at the time of the clinical examination; Alcohol, alcohol consumption grams per week; AP, suspected/confirmed angina pectoris; ASO, suspected/confirmed arteriosclerosis obliterans of lower limbs,; BMI, body mass index; CHD, suspected/confirmed congenital heart defect; CI, confidence interval; CVD, suspected/confirmed cerebral vascular disease; DBP, diastolic blood pressure; DM, diabetes mellitus; fS-chol, fasting serum cholesterol; fP-crea, fasting plasma creatinine; fP-gluc, fasting plasma glucose; fS-HDL-chol, fasting serum high-density lipoprotein cholesterol; fS-trigly, fasting serum triglycerides; HF, suspected/confirmed heart failure; HHD, suspected/confirmed hypertensive heart disease; HR, hazards ratio; HRa, heart rate; LBBB, left bundle branch block; MI, suspected/confirmed myocardial infarction; mv, multivariate; NCD, non-cardiac death; NSCD, non-sudden cardiac death; PHD, suspected/confirmed pulmonary heart disease; RBBB, right bundle branch block; SBP, systolic blood pressure; SCD, sudden cardiac death; uv, univariate; VHD, suspected/confirmed valvular heart disease. Please, see the Methods section for details. The clinical variables that differed significantly between the study participants with certain mode of death and those without such an event were included in the Cox mv regression analysis for corresponding death class group and the HRs in Table 3 are only shown for those selected clinical variables that remained significant in the Cox clinical hazards model. For the P-values lower than 0.000001, the P-values are shown as <0.000001; for the other P-values, the actual P-values are shown.
Table 4.
P-wave morphology parameters predicting sudden cardiac death, non-sudden cardiac death and non-cardiac death in the cox hazards model
| P-morphology parameters | SCD n = 379 | NSCD n = 928 | NCD n = 2440 | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P-value | HR | 95% CI | P-value | HR | 95% CI | P-value | ||
| P-dur | uv | 1.021 | 1.015–1.028 | <0.000001 | 1.024 | 1.020–1.029 | <0.000001 | 1.018 | 1.015–1.020 | <0.000001 |
| mv | 1.007 | 1.000–1.015 | 0.064 | 1.004 | 0.999–1.008 | 0.12 | 1.003 | 1.000–1.006 | 0.047 | |
| P-Max Amp-X | uv | 1.006 | 1.003–1.010 | 0.0004 | 1.006 | 1.004–1.008 | <0.000001 | 1.006 | 1.005–1.008 | <0.000001 |
| mv | 1.000 | 0.996–1.005 | 0.86 | 1.000 | 0.998–1.003 | 0.83 | 1.002 | 1.001–1.004 | 0.004 | |
| P-Max Amp-Y | uv | 1.002 | 1.000–1.005 | 0.11 | 0.999 | 0.997–1.000 | 0.09 | 1.000 | 0.999–1.001 | 0.69 |
| mv | 1.001 | 0.998–1.004 | 0.46 | 0.999 | 0.997–1.001 | 0.27 | 1.000 | 0.999–1.001 | 0.73 | |
| P-Max Amp-Z | uv | 1.028 | 1.024–1.032 | <0.000001 | 1.029 | 1.026–1.031 | <0.000001 | 1.021 | 1.020–1.023 | <0.000001 |
| mv | 1.010 | 1.005–1.015 | 0.0002 | 1.005 | 1.002–1.009 | 0.0005 | 1.002 | 1.000–1.005 | 0.03 | |
| P-PCA1 | uv | 0.997 | 0.996–0.998 | <0.000001 | 0.996 | 0.995–0.997 | <0.000001 | 0.998 | 0.997–0.998 | <0.000001 |
| mv | 0.999 | 0.999–1.000 | 0.21 | 1.000 | 0.999–1.000 | 0.19 | 1.000 | 0.9995–1.0002 | 0.45 | |
| P-PCA2 | uv | 1.022 | 1.016–1.028 | <0.000001 | 1.028 | 1.024–1.032 | <0.000001 | 1.020 | 1.017–1.022 | <0.000001 |
| mv | 1.003 | 0.997–1.009 | 0.34 | 1.003 | 0.999–1.007 | 0.17 | 1.002 | 1.000–1.004 | 0.11 | |
| P-PCA3 | uv | 1.033 | 1.019–1.047 | 0.000002 | 1.047 | 1.039–1.055 | <0.000001 | 1.034 | 1.029–1.040 | <0.000001 |
| mv | 1.008 | 0.994–1.023 | 0.27 | 1.012 | 1.002–1.021 | 0.01 | 1.008 | 1.002–1.014 | 0.01 | |
| PMD | uv | 1.016 | 1.010–1.021 | <0.000001 | 1.013 | 1.010–1.017 | <0.000001 | 1.010 | 1.008–1.012 | <0.000001 |
| mv | 1.006 | 1.000–1.012 | 0.05 | 0.999 | 0.995–1.003 | 0.68 | 0.999 | 0.997–1.001 | 0.44 | |
| PWH | uv | 1.018 | 1.007–1.030 | 0.002 | 1.009 | 1.000–1.017 | 0.04 | 1.009 | 1.004–1.014 | 0.0002 |
| mv | 1.006 | 0.993–1.018 | 0.38 | 1.007 | 0.999–1.015 | 0.09 | 1.008 | 1.003–1.013 | 0.002 | |
mv, multivariate; NCD, non-cardiac death; NSCD, non-sudden cardiac death; P-dur, P-wave duration; P-MaxAmp-X, the maximum positive amplitude of the P-wave in lead X; P-MaxAmp-Y, the maximum positive amplitude of the P-wave in lead Y; P-MaxAmp-Z, the maximum positive amplitude of the P-wave in lead Z; PMD, P-wave morphology dispersion; P-PCA1, the first dimension of the principal component analysis of the P-wave loop; P-PCA2, the second dimension of the principal component analysis of the P-wave loop; P-PCA3, the third dimension of the principal component analysis of the P-wave loop; PWH, P-wave heterogeneity; SCD, sudden cardiac death; uv, univariate. Please see the Methods section for details. For the P-values lower than 0.000001, the P-values are shown as <0.000001; for the other P-values, the actual P-values are shown.
The Kaplan–Meier curves show the cumulative proportional probabilities of SCD-free survival, NSCD-free survival, and NCD-free survival according to the dichotomized values of P-MaxAmp-Z and PMD (Figure 2 and Figure 3). When the significant clinical risk marker of SCD (Table 1) were added in the clinical risk model of SCD in the following order, the C-index increased as shown in parentheses: age (0.635), male gender (0.695), body mass index (0.700), heart rate (0.708), systolic blood pressure (0.716), diastolic blood pressure (0.716), diabetes (0.716), smoking (0.733), alcohol consumption (0.734), fasting plasma glucose (0.735), fasting serum cholesterol (0.748), fasting serum high-density lipoprotein cholesterol (0.748), fasting plasma creatinine (0.749), fasting serum triglycerides (0.749), right bundle branch block (0.749), myocardial infarction (0.759), angina pectoris (0.761), hypertension (0.761), heart failure (0.761), arteriosclerosis obliterans of lower limbs (0.762), PMD (as dichotomized variable; 0.764), and P-MaxAmp-Z (as dichotomized variable; 0.767). The association of P-wave morphology parameters with clinical risk markers are presented in Supplementary material online, Table S1 and Table S2. Most of the P-morphology parameters had a significant, but relatively weak, association with most of the clinical variables. An example of X, Y, Z lead ECG recording and the 12-lead ECG from a study participant who experienced SCD and from a study participant who remained alive during the follow-up is shown in Figure 4.
Figure 2.
The Kaplan–Meier curves showing the cumulative proportional probabilities of SCD-free survival, NSCD-free survival, and NCD-free survival for the study participants with the P-wave maximum amplitude in lead Z (P-MaxAmp-Z) ≥ 32.94 µV or < 32.94 µV. The cut-off points were optimized from the ROC curves. The statistical significance of the separation of the curves was assessed using the log-rank test.
Figure 3.
The Kaplan–Meier curves showing the cumulative proportional probabilities of SCD-free survival, NSCD-free survival, and NCD-free survival for the study participants with the PMD ≥ 44.45 degrees or < 44.45 degrees. The cut-off points were optimized from the ROC curves. The statistical significance of the separation of the curves was assessed using the log-rank test.
Figure 4.
An example of X, Y, and Z lead ECG recording and the 12-lead ECG from a study participant who experienced sudden cardiac death (above), (P-MaxAmp-Z = 123.0 µV, PMD = 69.7 degrees, PWH = 33.7) and from a study participant who remained alive (below), (P-MaxAmp-Z = 34.7 µV, PMD = 31.6 degrees, PWH = 25.0) during the follow-up. The abbreviations are the same as in Table 2.
Discussion
In the present large general population-based study with a long follow-up, most of the studied P-wave morphology parameters were associated with all modes of death in univariate comparisons, which is also illustrated in the Kaplan–Meier survival curves. A novel P-morphology parameter, PMD, had a close association with the risk of SCD but not with NSCD or NCD after relevant adjustments. The P-wave amplitude in the lead Z predicted SCD even in the multivariate analysis, however, also retained its association with other modes of death after adjustments. Adding PMD and P-MaxAmp-Z to the clinical risk model for SCD, increased the C-index slightly. However, some of the studied P-wave parameters, such as P-dur, P-MaxAmp-X, and PWH remained only associated with NCD, and PCA3 with NSCD and NCD in the multivariate analysis.
Changes in P-wave morphology are relatively commonly observed in electrocardiograms of middle-aged subjects.1 Abnormalities of P-wave have shown to yield prognostic information both in subjects with cardiovascular diseases and in general population. The studies have mainly focused on conventional measurements of P-wave, such P-terminal force, P-wave axis, and P-wave duration. P-terminal force has been found to predict cardiac mortality in postinfarction patients.9 In general population, P-terminal force has been shown to be associated with the risk of all-cause mortality, cardiovascular and ischaemic heart disease, atrial fibrillation, and SCD.1,5,6 P-wave axis4 and P-wave duration17 have also been observed to predict all-cause and cardiovascular mortality in general population. Prolongation of P-wave has been found to be associated with the occurrence of SCD in a general population.7 Our present study was focused on the evaluation of prognostic significance of novel P-wave morphology parameters analyzed from the orthogonal X-, Y-, Z-leads in the large general population during long follow-up. Our present observations are well in line with the previous observations based on the conventional measurements of P-wave from the 12-lead electrocardiogram supporting the concept that P-wave abnormalities yield also prognostic information in general population.1,4,5,6,17 Tereshchenko et al. have shown that deep terminal negativity of the P-wave in lead V1 in the 12-lead electrocardiogram predicts mortality and SCD in general population.5,6 These observations resemble the current findings with the maximum positive P-wave amplitude in the lead Z suggesting that these parameters may have a close relationship. Our present findings also suggest that some of the novel parameters describing spatial heterogeneity of atrial depolarization may have more specific association with the risk of SCD after relevant adjustments than parameters describing merely the duration or amplitude of atrial depolarization measured from a single lead.
Although explanations for the link between abnormalities of P-wave morphology and risk of death in general population remain speculative, there are several factors that may contribute to this association. P-wave abnormalities have been observed to be associated with left atrial enlargement, increased left ventricular end diastolic pressure, decreased left ventricular function,18 and elevated left atrial pressure.2 Increased left atrial pressure, left atrial enlargement, and increased left ventricular end diastolic pressure may indicate diastolic dysfunction that may be one link between P-wave abnormalities and risk of death. Abnormalities of P-wave have been shown to be associated with left ventricular interstitial fibrosis,3 which may favour the genesis of life-threatening ventricular arrhythmias. It has been shown in patients undergoing their first pulmonary vein isolation for persistent atrial fibrillation that P-wave prolongation is associated with atrial cardiomyopathy and left atrial hypertension.19 In our previous study, we have shown that abnormalities in P-wave morphology predict independently sudden cardiac death in patients with coronary artery disease.10 In that study novel P-morphology parameters were associated with soluble ST2, a marker of cardiac fibrosis, and with high-sensitivity C-reactive protein, a marker of inflammation. Inflammation is also considered a risk factor for SCD. Therefore, it is plausible to speculate that cardiac fibrosis and inflammation may also have contributed to the risk of SCD in the present general population. Changes in P-wave morphology are known to be a risk indicator for the development of atrial fibrillation,1 and there are previous data to show that atrial fibrillation is associated with the risk of SCD even in general population.20 Anatomical left ventricular hypertrophy is a well-known risk factor for SCD. Although we were unable to evaluate left ventricular hypertrophy in the present general population, it is probable that left ventricular hypertrophy was one of the links between abnormalities of P-wave morphology and SCD, a concept, which is supported by the observations of the association between P-wave changes and left ventricular mass index in our previous study10 and between P-wave changes and blood pressure in the present study. The explanation for the closer association of some of the P-morphology parameters with the risk of NCD than of cardiac death remains somewhat enigmatic. However, it is noteworthy that many P-wave morphology parameters were also associated with several clinical risk markers of NCD. It is also possible that in subjects with preponderant non-cardiac illness, increased cardiac risk profile represented by abnormalities of P-wave morphology may hasten the death.
The strengths of our study include a large study population and a long follow-up. Although the orthogonal electrocardiogram leads served as a robust platform for analyzing spatial heterogeneity of atrial depolarization, we were unable to measure the non-dipolar components of P-wave morphology as all the leads of the 12-lead electrocardiograms were not recorded simultaneously but in three lead sets. This also hindered the comparison of the diagnostic potency between the P-wave morphology parameters derived from the X-, Y-, Z-leads and the P-wave morphology parameters derived from the 12-lead ECG. In our previous study, we found that the non-dipolar components of P-wave morphology had the closest independent association with SCD in patients with coronary artery disease.10 One of the limitations of our study was that the ECGs were not initially digitalized, but were printed on paper, which reduces the diagnostic information, because the paper sweep speed and the line thickness of the ECG recorder limit the resolution of recording. Considering possible practical applications, it is noteworthy that the Frank X-, Y-, Z-leads can be derived from the standard 12-lead ECG.
In conclusion, P-wave morphology parameters yield long-term prognostic information in general population. Some of the novel parameters describing spatial heterogeneity of atrial depolarization have a close association with the risk of SCD and further improve the accuracy of the clinical model for predicting SCD in general population.
Supplementary material
Supplementary material is available at Europace online.
Supplementary Material
Contributor Information
Idamaria Laitinen, Research Unit of Internal Medicine, Medical Research Center Oulu, University of Oulu and Oulu University Hospital, Oulu, Finland.
Tuomas V Kenttä, Research Unit of Internal Medicine, Medical Research Center Oulu, University of Oulu and Oulu University Hospital, Oulu, Finland.
Jussi Passi, Research Unit of Internal Medicine, Medical Research Center Oulu, University of Oulu and Oulu University Hospital, Oulu, Finland.
Mira Anette E Haukilahti, Research Unit of Internal Medicine, Medical Research Center Oulu, University of Oulu and Oulu University Hospital, Oulu, Finland.
Antti Eranti, Heart Center, Turku University Hospital, Turku, Finland.
Arttu Holkeri, Division of Cardiology, Heart and Lung Center, Helsinki University Hospital and University of Helsinki, Helsinki, Finland.
Aapo L Aro, Division of Cardiology, Heart and Lung Center, Helsinki University Hospital and University of Helsinki, Helsinki, Finland.
Tuomas Kerola, Department of Internal Medicine, Päijät-Häme Central Hospital, Lahti, Finland.
Kai Noponen, Center for Machine Vision and Signal Analysis, University of Oulu, Oulu, Finland.
Tapio Seppänen, Center for Machine Vision and Signal Analysis, University of Oulu, Oulu, Finland.
Harri Rissanen, Finnish Institute for Health and Welfare, Helsinki, Finland.
Paul Knekt, Finnish Institute for Health and Welfare, Helsinki, Finland.
Markku Heliövaara, Finnish Institute for Health and Welfare, Helsinki, Finland.
Olavi H Ukkola, Research Unit of Internal Medicine, Medical Research Center Oulu, University of Oulu and Oulu University Hospital, Oulu, Finland.
M Juhani Junttila, Research Unit of Internal Medicine, Medical Research Center Oulu, University of Oulu and Oulu University Hospital, Oulu, Finland.
Heikki V Huikuri, Research Unit of Internal Medicine, Medical Research Center Oulu, University of Oulu and Oulu University Hospital, Oulu, Finland.
Juha S Perkiömäki, Research Unit of Internal Medicine, Medical Research Center Oulu, University of Oulu and Oulu University Hospital, Oulu, Finland.
Funding
The study is supported by a grant from the Ida Montin Foundation, Helsinki, Finland, the University of Oulu Scholarship Foundation, Oulu, Finland, the Finnish Foundation for Cardiovascular Research, Helsinki, Finland
Data availability
The data underlying this article were provided by permission. The data will be shared on reasonable request to the corresponding author and the Finnish Institute for Health and Welfare/Findata.
References
- 1. Eranti A, Aro AL, Kerola T, Anttonen O, Rissanen HA, Tikkanen JT, et al. Prevalence and prognostic significance of abnormal P terminal force in lead V1 of the ECG in the general population. Circ Arrhythm Electrophysiol 2014;7:1116–21. [DOI] [PubMed] [Google Scholar]
- 2. Petersson R, Smith JG, Larsson DA, Reitan Ö, Carlson J, Platonov P, et al. Orthogonal P-wave morphology is affected by intra-atrial pressures. BMC Cardiovasc Disord 2017;17:288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Win TT, Venkatesh BA, Volpe GJ, Mewton N, Rizzi P, Sharma RK, et al. Associations of electrocardiographic P-wave characteristics with left atrial function, and diffuse left ventricular fibrosis defined by cardiac magnetic resonance: The PRIMERI Study. Heart Rhythm 2015;12:155–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Li Y, Shah AJ, Soliman EZ. Effect of electrocardiographic P-wave axis on mortality. Am J Cardiol 2014;113:372–6. [DOI] [PubMed] [Google Scholar]
- 5. Tereshchenko LG, Shah AJ, Li Y, Soliman EZ. Electrocardiographic deep terminal negativity of the P wave in V1 and risk of mortality: The national health and nutrition examination survey III. J Cardiovasc Electrophysiol 2014;25:1242–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Tereshchenko LG, Henrikson CA, Sotoodehnia N, Arking DE, Agarwal SK, Siscovick DS, et al. Electrocardiographic deep terminal negativity of the P wave in V1 and risk of sudden cardiac death: the atherosclerosis risk in communities (ARIC) study. J Am Heart Assoc 2014;3:e001387. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Maheshwari A, Norby FL, Soliman EZ, Alraies MC, Adabag S, O’Neal WT, et al. Relation of prolonged P-wave duration to risk of sudden cardiac death in the general population (from the Atherosclerosis risk in communities study). Am J Cardiol 2017;119:1302–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Eranti A, Carlson J, Kenttä T, Holmqvist F, Holkeri A, Haukilahti MA, et al. Orthogonal P-wave morphology, conventional P-wave indices, and the risk of atrial fibrillation in the general population using data from the Finnish hospital discharge register. Europace 2020;22:1173–81. [DOI] [PubMed] [Google Scholar]
- 9. Perkiomaki JS, Zareba W, Greenberg HM, Moss AJ, Thrombogenic Factors and Recurrent Coronary Events Investigators . Usefulness of standard electrocardiographic parameters for predicting cardiac events after acute myocardial infarction during modern treatment era. Am J Cardiol 2002;90:205–9. [DOI] [PubMed] [Google Scholar]
- 10. Nortamo S, Laitinen I, Passi J, Tulppo M, Ukkola OH, Junttila J, et al. Prognostic significance of P-wave morphology in patients with coronary artery disease. J Cardiovasc Electrophysiol 2019;30:2051–60. [DOI] [PubMed] [Google Scholar]
- 11. Mini-Finland Health Survey . Finnish institute for health and welfare. https://thl.fi/en/web/thlfi-en/research-and-expertwork/population-studies/finnish-mobile-clinic/mini-finland-health-survey
- 12. Knekt P, Rissanen H, Järvinen R, Heliövaara M. Cohort profile: the Finnish mobile clinic health survey FMC, FMCF and MFS. Int J Epidemiol 2017;46:1760–1. [DOI] [PubMed] [Google Scholar]
- 13. Pratt CM, Greenway PS, Schoenfeld MH, Hibben ML, Reiffel JA. Exploration of the precision of classifying sudden cardiac death. Implications for the interpretation of clinical trials. Circulation 1996;93:519–24. [DOI] [PubMed] [Google Scholar]
- 14. Holkeri A, Eranti A, Kenttä TV, Noponen K, Haukilahti MAE, Seppänen T, et al. Experiences in digitizing and digitally measuring a paper-based ECG archive. J Electrocardiol 2018;51:74–81. [DOI] [PubMed] [Google Scholar]
- 15. Nearing BD, Verrier RL. Tracking cardiac electrical instability by computing interlead heterogeneity of T-wave morphology. J Appl Physiol 2003;95:2265–72. [DOI] [PubMed] [Google Scholar]
- 16. Acar B, Yi G, Hnatkova K, Malik M. Spatial, temporal and wavefront direction characteristics of 12-lead T-wave morphology. Med Biol Eng Comput 1999;37:574–84. [DOI] [PubMed] [Google Scholar]
- 17. Magnani JW, Gorodeski EZ, Johnson VM, Sullivan LM, Hamburg NM, Benjamin EJ, et al. P wave duration is associated with cardiovascular and all-cause mortality outcomes: the national health and nutrition examination survey. Heart Rhythm 2011;8:93–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Mehta A, Jain AC, Mehta MC, Billie M. Left atrial abnormality in acute myocardial infarction. Am J Cardiol 1997;79:807–11. [DOI] [PubMed] [Google Scholar]
- 19. Eichenlaub M, Mueller-Edenborn B, Minners J, Jander N, Allgeier M, Lehrmann H, et al. Left atrial hypertension, electrical conduction slowing, and mechanical dysfunction - the pathophysiological triad in atrial fibrillation-associated atrial cardiomyopathy. Front Physiol 2021;12:670527. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Chen LY, Sotoodehnia N, Bůžková P, Lopez FL, Yee LM, Heckbert SR, et al. Atrial fibrillation and the risk of sudden cardiac death: the atherosclerosis risk in communities study and cardiovascular health study. JAMA Intern Med 2013;173:29–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article were provided by permission. The data will be shared on reasonable request to the corresponding author and the Finnish Institute for Health and Welfare/Findata.