2024 Volume 31 Issue 3 Pages 214-231
Aim: Sudden death is one of the most common causes of death among hemodialysis patients. Electrocardiography (ECG) is a noninvasive and inexpensive test that is regularly performed in hemodialysis clinics. However, the association between abnormal ECG findings and the risk of sudden death in hemodialysis patients is yet to be fully elucidated. Thus, the aim of this study was to determine the ECG parameters linked to sudden death in patients undergoing hemodialysis.
Methods: The Q-Cohort Study is a multicenter, longitudinal, observational study of hemodialysis patients. In this study, 1,153 Japanese hemodialysis patients aged ≥ 18 years with ECG data recorded within 1 year of study enrollment were followed up for 10 years. Cox proportional hazards models were used to estimate the multivariate-adjusted hazard ratios (HRs) with 95% confidence intervals (CIs) for the association between ECG parameters and sudden death.
Results: During the median follow-up period of 9.0 years, 517 patients died, 76 of whom exhibited sudden death. After adjusting for confounding factors, higher heart rate, QT prolongation, and left ventricular hypertrophy as per the Sokolow–Lyon voltage criteria were found to be independently associated with an increased risk of sudden death. The adjusted HRs [95% CIs] for each abnormal ECG parameter were 2.02 [1.05–3.89], 2.10 [1.30–1.77], and 1.91 [1.18–3.09], respectively.
Conclusions: Higher heart rate, QT prolongation, and left ventricular hypertrophy on ECG have been determined to be associated with an increased risk of sudden death. Therefore, regular ECG recording could enable medical practitioners to identify hemodialysis patients who require intervention to prevent lethal arrhythmia.
Cardiovascular disease has been identified as the most common cause of death among patients undergoing hemodialysis1). Sudden death, which is defined as sudden and unexpected natural death within a short period (generally between 1 and 24 hours) after symptom onset, is a serious concern that is often observed in patients undergoing maintenance hemodialysis2). Importantly, the incidence of sudden death is noted to be significantly higher in patients undergoing hemodialysis than in the general population3). According to the United States Renal Data System, sudden death is the leading cause of death in dialysis patients, accounting for approximately 40% of all known deaths1). Similarly, we reported that 13% of all deaths in Japanese patients undergoing hemodialysis were caused by sudden death4). In the general population, the main cause of sudden death is presumed to be fatal arrhythmias induced by myocardial ischemia, usually in the presence of coronary artery disease5, 6). In contrast, in hemodialysis patients, the detailed mechanisms of sudden death are yet to be fully understood, although a variety of risk factors, such as left ventricular hypertrophy (LVH), myocardial fibrosis, rapid fluctuations in fluid volume and electrolytes, and sympathetic overactivity, have been proposed7). Early detection of risk factors for sudden death and timely intervention are clinically important to prevent sudden death in this population.
Electrocardiography (ECG) is a noninvasive and inexpensive test to detect signs of cardiovascular disease. In Japan, ECG is periodically performed on hemodialysis patients as a routine test before hemodialysis sessions every 1–3 months, even if patients do not complain of any symptoms. Several studies have reported an association between abnormal ECG findings and prognosis in patients undergoing hemodialysis5, 8). However, the relationship between ECG parameters and sudden death in hemodialysis patients is yet to be fully understood.
In this study, we aimed to determine the association between ECG parameters and the risk of sudden death in patients undergoing maintenance hemodialysis using the data from the Q-Cohort Study, which is a multicenter, longitudinal, observational study that examined Japanese patients undergoing maintenance hemodialysis9).
The Q-Cohort Study is a multicenter, longitudinal, observational study of patients undergoing hemodialysis in Japan. The details of this study have been previously described9, 10). In total, 3,598 outpatients aged ≥ 18 years who underwent maintenance hemodialysis at 39 dialysis facilities in the Fukuoka and Saga Prefectures in the northern region of Kyushu were enrolled in this study from December 2006 to December 2007. The participants were followed up until December 2016. The patients who were lost to follow-up during the follow-up period were regarded as “censored” on the day of the final visit and were included in the analyses. After the 10-year follow-up period, we retrospectively collected ECG data recorded within 1 year from the study enrollment. Although ECG was performed regularly at all facilities, the ECG data at baseline (from 2006 to 2007) had already been discarded for 2,379 patients. The data for the remaining 1,219 patients were available. After excluding patients without baseline demographic data (n=7) and patients with atrial fibrillation (n=48) or post-pacemaker implantation (n=11), we included the remaining 1,153 patients in this study. A flow chart of patient inclusion is shown in Fig.1. The study protocol was approved by the Clinical Research Ethics Committee of the Institutional Review Board of Kyushu University (approval number: 20-31) and was registered in the clinical trial registry (University Hospital Medical Information Network, UMIN000000556). Written informed consent was obtained from all patients before their participation in the study. This study was performed in accordance with the principles of the Declaration of Helsinki. The ethics committees of all participating institutions waived the requirement for written informed consent for the additional follow-up survey from 2011 to 2016 because of the retrospective nature of this study.
PMI, pacemaker implantation.
Resting 12-lead ECG was performed just before hemodialysis sessions. The ECG parameters evaluated in this study are shown in Fig.2. In this study, we selected ECG parameters from the viewpoint of automatic measurement. We targeted five parameters, which were automatically measured by ECG device and listed on the sheet, including the RR interval (heart rate), PR interval, QRS duration, QT interval, and LVH as per the Sokolow–Lyon criteria. Using the RR interval, patients were divided into quartiles (Q1–Q4) according to the baseline heart rate as follows: Q1 (n=279), ＜64 bpm; Q2 (n=279), 65–72 bpm; Q3 (n=283), 73–80 bpm; and Q4 (n=312), ≥ 81 bpm. The PR interval extended from the onset of the P wave until the end of the PR segment (junction with the QRS complex). We defined PR interval prolongation as a PR interval ≥ 220 msec, as reported previously11). The QRS duration was from the onset to the termination of the QRS wave. QRS duration prolongation was defined ≥ 110 mecs, as previously described12). The QT interval was measured from the onset of the QRS complex to the end of the T wave. The QT interval was corrected for each patient’s heart rate using Bazett’s formula: QTc=QT÷√RR msec13). QTc prolongation was defined as QTc ≥ 460 msec in women and QTc ≥ 450 msec in men, as previously reported14). The Sokolow–Lyon criteria were used to identify LVH by ECG. The cut-off point for LVH by Sokolow–Lyon voltage (RV5/6＋SV1) was ≥ 35 mm. Correlation coefficients between each ECG parameter were calculated. Along with these five parameters, the JT interval, which is proposed as a more appropriate measure of ventricular repolarization than the QT interval, was also evaluated. The corrected JT (JTc) interval was defined as QTc−QRS duration. As the cut-off value of the JTc interval has not been established previously, we investigated the cut-off value of JTc interval for the prediction of sudden death using receiver operating characteristic (ROC) curve and area under the curve (AUC) analyses. JTc interval was divided into two groups by the cut-off value, and JTc prolongation was defined as a group with a higher JTc interval value than the cut-off value in this study.
(a) Representative illustration of each ECG parameter. (b) Definition of SV1 and RV5 for LVH according to the Sokolow–Lyon voltage criteria. If SV1＋RV5/6 ≥ 35 mm, the patients were defined as having LVH. ECG, electrocardiography; LVH, left ventricular hypertrophy.
The primary outcome was sudden death, which is defined as witnessed death within 24 hours of the onset of acute symptoms or unwitnessed unexpected death within the interval between dialysis sessions, excluding death from trauma, suicide, and suffocation. This definition has been used previously15). Patients’ health statuses were checked annually by local physicians at each dialysis facility. We collected death events from medical records or death certificates by visiting each hemodialysis facility. The data obtained for each deceased patient were reviewed by the physician members of the Q-Cohort Study to confirm the cause of death. For patients who were enrolled in this study but were transferred to other hemodialysis facilities, we inquired about the patients’ medical condition via telephone or mail. Demographic information (e.g., age, sex, diabetic nephropathy, history of cardiovascular events, and duration of hemodialysis), serum biochemical data (e.g., serum albumin, calcium, phosphate, and total cholesterol concentration), and medication details at baseline were obtained from medical records. Blood samples were collected from vascular access before dialysis at the beginning of the week. The corrected serum calcium concentration was calculated using Payne’s formula as follows: corrected serum calcium concentration (mg/dL)=observed serum calcium concentration (mg/dL)＋(4.0−serum albumin concentration [g/dL])16). This calculation was applied only when the serum albumin concentration was ＜4 g/dL.
Statistical AnalysisBaseline patient data are presented as median (interquartile range) for continuous variables or number (percentage) for categorical measures. PR prolongation, QRS prolongation, QTc prolongation, and LVH were compared between the groups using the Mann–Whitney U test for continuous variables and the chi-square test for categorical variables. Trends among the quartiles of baseline heart rate were examined by the Cochran–Armitage test for categorical parameters and by the Jonckheere–Terpstra test for continuous parameters. Simple correlation coefficients between each ECG parameter were analyzed using Pearson’s correlation coefficient. Multivariate-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for sudden death were calculated using the Cox proportional hazards model. Model 1 was unadjusted. Model 2 was adjusted for age and sex. Model 3 was adjusted for the variables in Model 2 plus the presence of diabetic nephropathy, history of cardiovascular events, duration of hemodialysis, and systolic blood pressure. Model 4 was adjusted for the variables in Model 3 plus the serum albumin, total cholesterol, corrected calcium, and phosphate concentrations. The multivariate-adjusted model included the following potential confounding factors based on a priori clinical judgment: age, sex, presence of diabetic nephropathy, history of cardiovascular events, duration of hemodialysis, systolic blood pressure, serum albumin, total cholesterol, corrected calcium, and phosphate concentrations. We have also applied a restricted cubic spline curve model to analyze the association between baseline ECG parameters such as a continuous variable and the risk of sudden death. As a sensitivity analysis, the associations between the abnormal ECG parameters and the risk of sudden death were examined. The abnormal ECG findings included higher heart rate (≥ 80 bpm), QT prolongation, and LVH. Subgroup analyses were performed to estimate heterogeneity according to the subgroups of risk factors. Variables relevant to the subgroups were excluded from each model. A two-tailed P-value of ＜0.05 was considered to be statistically significant for all analyses. Statistical analyses were performed using R statistical software, version 3.0.2 (http://cran.rproject.org).
The baseline clinical characteristics of our cohort are shown in Table 1. The median age of the patients was 62.6 years, and 61.2% of the patients were male. The median duration of dialysis was 7.8 years. The proportion of patients with diabetic nephropathy was 29%. The prevalence or history of cardiovascular disease was 18%. The median predialysis systolic blood pressure was 155 mmHg. The median serum albumin, total cholesterol, corrected calcium, and phosphate concentrations were 164 mg/dL, 4.0 g/dL, 9.3 mg/dL, and 4.8 mg/dL, respectively. The clinical background characteristics of the excluded patients (n=2,445) and the included patients (n=1,153) are shown in Supplemental Table 1. The patients included in this study were noted to be younger (P=0.003) and had a lower prevalence of cardiovascular disease (P＜0.001) than the excluded patients. The included patients also had a higher body mass index (P=0.001), Kt/V for urea (P＜0.001), systolic blood pressure (P＜0.001), serum albumin concentration (P＜0.001), and serum total cholesterol concentration (P＜0.001), as well as a lower cardiothoracic ratio (P＜0.001), serum C-reactive protein concentration (P＜0.001), corrected serum calcium concentration (P＜0.001), and serum phosphate concentration (P＜0.001) than the excluded patients. The frequency of phosphate binder use was significantly higher for the included patients than the excluded patients (P＜0.001). The baseline clinical characteristics for each ECG finding are shown in Supplementary Tables 2, 3, 4, 5, 6.
| Parameters | Summary data |
|---|---|
| Demographics and dialysis-related information | |
| Age, years | 62.6 (55.1–72.0) |
| Male sex, n (%) | 706 (61.2) |
| Duration of dialysis, years | 7.8 (2.4–11.3) |
| Diabetic nephropathy, n (%) | 334 (29.0) |
| History of cardiovascular diseases, n (%) | 208 (18.0) |
| Body mass index, kg/m2 | 21.4 (19.6–22.6) |
| Kt/V for urea | 1.56 (1.43–1.74) |
| nPCR, g/kg/day | 0.95 (0.85–1.05) |
| Systolic blood pressure, mmHg | 155 (140–168) |
| Cardiothoracic ratio, % | 49.7 (46.2–53.3) |
| Blood tests | |
| Hemoglobin, g/dL | 10.6 (9.9–11.2) |
| Serum albumin, g/dL | 4.0 (3.7–4.2) |
| Serum total cholesterol, mg/dL | 164 (135–182) |
| Serum creatinine, mg/dL | 10.1 (8.5–12.2) |
| Serum C-reactive protein, mg/dL | 0.13 (0.05–0.27) |
| Corrected serum calcium, mg/dL | 9.3 (8.9–9.8) |
| Serum phosphate, mg/dL | 4.8 (4.1–5.5) |
| Serum PTH (intact assay), pg/mL | 107.1 (48.1–211) |
| Medications | |
| Use of anti-hypertensives, n (%) | 433 (37.6) |
| Use of ESAs, n (%) | 962 (83.4) |
| Use of phosphate binders, n (%) | 1005 (87.1) |
| Use of VDRAs, n (%) | 821 (71.2) |
The data are shown as the median (interquartile range) or number (percentage). ESA, erythropoiesis-stimulating agent; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone; VDRA, vitamin D receptor activator.
| Parameters | ECG not available n = 2445 | ECG available n = 1153 | P-value |
|---|---|---|---|
| Demographics and dialysis-related information | |||
| Age, years | 64.7 (56.2-73.2) | 63.4 (55.1-72.0) | 0.003 |
| Male sex, (%) | 1388 (58.4) | 707 (61.3) | 0.11 |
| Dialysis vintage, years | 5.0 (2.0-11.1) | 5.8 (2.4-11.3) | 0.08 |
| Diabetic nephropathy, (%) | 700 (29.5) | 335 (29.1) | 0.76 |
| History of cardiovascular diseases, (%) | 578 (24.3) | 209 (18.1) | ＜0.001 |
| Body mass index, kg/m2 | 20.7 (19.0-22.7) | 21.4 (19.6-22.6) | 0.001 |
| Kt/V for urea | 1.56 (1.41-1.69) | 1.56 (1.43-1.74) | 0.009 |
| nPCR, g/kg/day | 0.95 (0.86-1.04) | 0.95 (0.85-1.05) | 0.63 |
| Systolic blood pressure, mmHg | 152 (137-168) | 156 (140-168) | ＜0.001 |
| Cardiothoracic ratio, (%) | 50.0 (45.0-53.9) | 49.7 (46.1-53.3) | ＜0.001 |
| Blood tests | |||
| Hemoglobin, g/dL | 10.5 (9.8-11.3) | 10.6 (9.9-11.2) | 0.39 |
| Serum albumin, g/dL | 3.8 (3.7-4.2) | 4.0 (3.7-4.2) | ＜0.001 |
| Serum total cholesterol, mg/dL | 150 (128-176) | 157 (135-182) | ＜0.001 |
| Serum creatinine, mg/dL | 10.1 (8.3-11.9) | 10.1 (8.5-12.2) | 0.13 |
| Serum C-reactive protein, mg/dL | 0.13 (0.06-0.33) | 0.13 (0.05-0.27) | ＜0.001 |
| Serum corrected serum calcium, mg/dL | 9.4 (8.9-9.9) | 9.3 (8.9-9.8) | ＜0.001 |
| Serum phosphate, mg/dL | 4.9 (4.2-5.7) | 4.8 (4.1-5.5) | 0.006 |
| Serum PTH (intact assay), pg/mL | 106 (48-218) | 108 (48-211) | 0.85 |
| Medications | |||
| Use of anti-hypertensives, (%) | 1490 (62.7) | 716 (62.1) | 0.74 |
| Use of ESAs, (%) | 2008 (84.5) | 963 (83.5) | 0.33 |
| Use of phosphate binders, (%) | 1853 (78.0) | 1006 (87.2) | ＜0.001 |
| Use of VDRAs, (%) | 1643 (69.2) | 822 (71.3) | 0.22 |
Data are shown as median (interquartile range) or number (percentage). ESA, erythropoiesis stimulating agent; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone; VDRA, vitamin D receptor activator.
| Quartiles of baseline heart rate, bpm | |||||
|---|---|---|---|---|---|
|
Q1 (≤ 64), n = 279 |
Q2 (65–72) n = 279 |
Q3 (73–80) n = 283 |
Q4 ( ≥81 bpm) n = 312 |
P-for trend | |
| Demographics and dialysis-related information | |||||
| Age, years | 66.3 (57.8–74.0) | 64.4 (56.0–72.2) | 62.0 (55.4–71.9) | 60.1 (52.2–69.7) | ＜0.001 |
| Male sex, n (%) | 171 (61.3) | 173 (62.0) | 163 (57.6) | 199 (63.8) | 0.77 |
| Duration of dialysis, years | 4.8 (1.8–9.8) | 5.8 (2.4–12.0) | 6.1 (2.9–11.3) | 6.3 (2.3–12.0) | 0.03 |
| Diabetic nephropathy, n (%) | 73 (26.2) | 80 (28.7) | 85 (30.0) | 96 (30.8) | 0.20 |
| History of cardiovascular diseases, n (%) | 58 (20.8) | 43 (15.4) | 51 (18.0) | 56 (17.9) | 0.58 |
| Body mass index, kg/m2 | 20.8 (19.6–22.0) | 21.2 (19.7–22.4) | 21.4 (19.5–23.1) | 21.5 (19.9–23.1) | 0.01 |
| Kt/V for urea | 1.56 (1.43–1.74) | 1.56 (1.48–1.77) | 1.56 (1.42–1.75) | 1.56 (1.40–1.69) | 0.03 |
| nPCR, g/kg/day | 0.95 (0.86–1.05) | 0.95 (0.86–1.05) | 0.95 (0.84–1.04) | 0.95 (0.84–1.03) | 0.18 |
| Systolic blood pressure, mmHg | 156 (142–168) | 152 (140–168) | 156 (141–168) | 156 (139–169) | 0.94 |
| Cardiothoracic ratio, % | 50.4 (46.5–53.3) | 49.4 (46.1–52.9) | 50.0 (46.8–53.3) | 49.4 (45.6–53.8) | 0.21 |
| Blood tests | |||||
| Hemoglobin, g/dL | 10.6 (9.9–11.2) | 10.5 (9.9–11.1) | 10.6 (9.9–11.3) | 10.6 (9.9–11.2) | 0.58 |
| Serum albumin, g/dL | 4.0 (3.8–4.2) | 3.9 (3.7–4.2) | 4.0 (3.7–4.2) | 3.9 (3.7–4.2) | 0.52 |
| Serum total cholesterol, mg/dL | 158 (133–180) | 156 (135–180) | 157 (135–185) | 158 (138–179) | 0.30 |
| Serum creatinine, mg/dL | 9.8 (8.45–11.9) | 9.9 (8.4–11.9) | 10.6 (8.7–12.6) | 10.1 (8.5–12.4) | 0.07 |
| Serum C-reactive protein, mg/dL | 0.11 (0.05–0.20) | 0.12 (0.05–0.26) | 0.13 (0.05–0.27) | 0.13 (0.06–0.35) | 0.008 |
| Corrected serum calcium, mg/dL | 9.3 (8.85–9.7) | 9.3 (8.8–9.8) | 9.3 (8.9–9.8) | 9.4 (8.9–9.9) | 0.10 |
| Serum phosphate, mg/dL | 4.8 (4.1–5.5) | 4.7 (4.0–5.5) | 4.9 (4.2–5.6) | 4.8 (4.1–5.6) | 0.34 |
| Serum PTH (intact assay), pg/mL | 103 (46–208) | 110 (46–200) | 106 (49–228) | 113 (55–210) | 0.42 |
| Medications | |||||
| Use of anti-hypertensives, n (%) | 192 (68.8) | 173 (62.0) | 172 (60.8) | 178 (57.1) | 0.005 |
| Use of ESAs, n (%) | 250 (89.6) | 230 (82.4) | 234 (82.7) | 248 (79.5) | 0.002 |
| Use of phosphate binders, n (%) | 251 (90.0) | 237 (84.9) | 251 (88.7) | 266 (85.3) | 0.23 |
| Use of VDRAs, n (%) | 209 (74.9) | 201 (72.0) | 193 (68.2) | 218 (69.9) | 0.12 |
Data are shown as median (interquartile range) or number (percentage). ESA, erythropoiesis-stimulating agent; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone; VDRA, vitamin D receptor activator; Q, quartile of heart rate.
| Parameters | PR prolongation (−) n = 101 | PR prolongation (+) n = 1052 | P-value |
|---|---|---|---|
| Demographics and dialysis-related information | |||
| Age, years | 62.7 (54.8–71.3) | 70.8 (63.3–77.3) | ＜0.001 |
| Male sex, n (%) | 628 (59.7) | 78 (77.2) | 0.001 |
| Duration of dialysis, years | 5.8 (2.3–11.4) | 6.3 (2.7–9.8) | 0.99 |
| Diabetic nephropathy, n (%) | 290 (27.6) | 44 (43.6) | ＜0.001 |
| History of cardiovascular diseases, n (%) | 171 (16.3) | 37 (36.6) | ＜0.001 |
| Body mass index, kg/m2 | 21.4 (19.6–22.7) | 21.5 (20.2–22.0) | 0.84 |
| Kt/V for urea | 1.56 (1.43–1.74) | 1.54 (1.39–1.67) | 0.01 |
| nPCR, g/kg/day | 0.95 (0.86–1.05) | 0.95 (0.79–1.00) | 0.02 |
| Systolic blood pressure, mmHg | 156 (140–168) | 157 (140–168) | 0.84 |
| Cardiothoracic ratio, % | 49.5 (46.0–53.1) | 51.1 (48.4–55.0) | ＜0.001 |
| Blood tests | |||
| Hemoglobin, g/dL | 10.6 (9.9–11.2) | 10.4 (9.8–11.2) | 0.39 |
| Serum albumin, g/dL | 4.0 (3.7–4.2) | 3.9 (3.0–4.2) | 0.20 |
| Serum total cholesterol, mg/dL | 158 (136–183) | 147 (130–169) | 0.00 |
| Serum creatinine, mg/dL | 10.1 (8.5–12.2) | 9.7 (8.0–11.5) | 0.03 |
| Serum C-reactive protein, mg/dL | 0.13 (0.05–0.25) | 0.14 (0.07–0.40) | 0.04 |
| Corrected serum calcium, mg/dL | 9.3 (8.9–9.8) | 9.3 (8.8–9.8) | 0.92 |
| Serum phosphate, mg/dL | 4.8 (4.1–5.6) | 4.7 (3.9–5.4) | 0.26 |
| Serum PTH (intact assay), pg/mL | 110 (49–212) | 93 (46–176) | 0.28 |
| Medications | |||
| Use of anti-hypertensives, n (%) | 649 (61.7) | 66 (65.3) | 0.30 |
| Use of ESAs, n (%) | 876 (83.3) | 86 (85.1) | 0.78 |
| Use of phosphate binders, n (%) | 916 (87.1) | 89 (88.1) | 0.88 |
| Use of VDRAs, n (%) | 747 (71.0) | 74 (73.3) | 0.73 |
Data are shown as median (interquartile range) or number (percentage). ESA, erythropoiesis-stimulating agent; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone; VDRA, vitamin D receptor activator.
| Parameters | QRS prolongation (−) n = 233 | QRS prolongation (+) n = 920 | P-value |
|---|---|---|---|
| Demographics and dialysis-related information | |||
| Age, years | 62.7 (54.7–71.3) | 66.7 (57.9–73.9) | ＜0.001 |
| Male sex, (%) | 531 (57.8) | 175 (75.1) | ＜0.001 |
| Duration of dialysis, years | 5.9 (2.4–11.5) | 4.8 (2.3–9.8) | 0.14 |
| Diabetic nephropathy, (%) | 253 (27.5) | 81 (34.8) | 0.04 |
| History of cardiovascular diseases, (%) | 135 (14.7) | 72 (30.9) | ＜0.001 |
| Body mass index, kg/m2 | 21.2 (19.6–22.6) | 21.5 (19.9–22.7) | 0.103 |
| Kt/V for urea | 1.56 (1.44–1.75) | 1.52 (1.36–1.68) | ＜0.001 |
| nPCR, g/kg/day | 0.95 (0.86–1.05) | 0.95 (0.83–1.04) | 0.09 |
| Systolic blood pressure, mmHg | 154 (140–168) | 160 (144–170) | 0.07 |
| Cardiothoracic ratio, % | 49.5 (46.0–53.0) | 50.5 (46.7–54.3) | 0.04 |
| Blood tests | |||
| Hemoglobin, g/dL | 10.6 (9.9–11.2) | 10.6 (9.9–11.3) | 0.55 |
| Serum albumin, g/dL | 4.0 (3.7–4.2) | 4.0 (3.7–4.2) | 0.52 |
| Serum total cholesterol, mg/dL | 158 (136–182) | 154 (134–182) | 0.28 |
| Serum creatinine, mg/dL | 10.1 (8.5–12.2) | 10.0 (8.5–12.0) | 0.51 |
| Serum C-reactive protein, mg/dL | 0.13 (0.05–0.25) | 0.13 (0.06–0.34) | 0.04 |
| Corrected serum calcium, mg/dL | 9.3 (8.8–9.8) | 9.3 (8.9–9.9) | 0.34 |
| Serum phosphate, mg/dL | 4.8 (4.1–5.5) | 4.9 (4.0–5.6) | 0.76 |
| Serum PTH (intact assay), pg/mL | 108 (49–211) | 104 (48–223) | 0.78 |
| Medications | |||
| Use of anti-hypertensives, n (%) | 556 (60.5) | 158 (67.8) | 0.10 |
| Use of ESAs, n (%) | 767 (83.5) | 194 (83.3) | 0.92 |
| Use of phosphate binders, n (%) | 791 (86.1) | 214 (91.8) | 0.02 |
| Use of VDRAs, n (%) | 660 (71.8) | 160 (68.7) | 0.37 |
Data are shown as median (interquartile range) or number (percentage). ESA, erythropoiesis-stimulating agent; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone; VDRA, vitamin D receptor activator.
| Parameters | QTc prolongation (−) n = 250 | QTc prolongation (+) n = 903 | P-value |
|---|---|---|---|
| Demographics and dialysis-related information | |||
| Age, years | 62.6 (54.7–71.5) | 66.3 (57.7–73.0) | ＜0.001 |
| Male sex, n (%) | 529 (58.6) | 177 (70.8) | ＜0.001 |
| Duration of dialysis, years | 6.0 (2.5–12.1) | 4.4 (1.9–9.6) | 0.009 |
| Diabetic nephropathy, n (%) | 230 (25.5) | 104 (41.6) | ＜0.001 |
| History of cardiovascular diseases, n (%) | 137 (15.2) | 71 (28.4) | ＜0.001 |
| Body mass index, kg/m2 | 21.2 (19.6–22.6) | 21.5 (19.9–22.7) | 0.12 |
| Kt/V for urea | 1.56 (1.44–1.76) | 1.56 (1.38–1.63) | ＜0.001 |
| nPCR, g/kg/day | 0.95 (0.86–1.05) | 0.95 (0.83–1.01) | 0.005 |
| Systolic blood pressure, mmHg | 154 (140–168) | 160 (144–174) | 0.002 |
| Cardiothoracic ratio, % | 49.6 (46.0–53.0) | 50.0 (46.6–54) | 0.14 |
| Blood tests | |||
| Hemoglobin, g/dL | 10.6 (9.9–11.2) | 10.6 (9.8–11.1) | 0.16 |
| Serum albumin, g/dL | 4.0 (3.7–4.2) | 3.9 (3.7–4.2) | 0.45 |
| Serum total cholesterol, mg/dL | 154 (144–168) | 160 (144–174) | 0.09 |
| Serum creatinine, mg/dL | 10.3 (8.5–12.4) | 9.6 (8.1–11.6) | 0.002 |
| Serum C-reactive protein, mg/dL | 0.12 (0.05–0.22) | 0.15 (0.06–0.43) | ＜0.001 |
| Corrected serum calcium, mg/dL | 9.4 (8.9–9.8) | 9.1 (8.8–9.8) | 0.005 |
| Serum phosphate, mg/dL | 4.9 (4.1–5.6) | 4.6 (3.9–5.5) | 0.09 |
| Serum PTH (intact assay), pg/mL | 109 (50–216) | 104 (46–190) | 0.49 |
| Medications | |||
| Use of anti-hypertensives, n (%) | 552 (61.1) | 163 (65.2) | 0.45 |
| Use of ESAs, n (%) | 740 (81.9) | 222 (88.8) | 0.01 |
| Use of phosphate binders, n (%) | 789 (87.4) | 216 (86.4) | 0.67 |
| Use of VDRAs, n (%) | 657 (72.8) | 164 (65.6) | 0.03 |
Data are shown as median (interquartile range) or number (percentage). ESA, erythropoiesis-stimulating agent; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone; VDRA, vitamin D receptor activator.
| Parameters | LVH (−) n = 251 | LVH (+) n = 902 | P-value |
|---|---|---|---|
| Demographics and dialysis-related information | |||
| Age, years | 63.3 (54.7–71.9) | 63.9 (57.2–72.4) | 0.27 |
| Male sex, n (%) | 548 (60.8) | 158 (62.9) | 0.56 |
| Duration of dialysis, years | 5.8 (2.4–11.4) | 5.9 (2.3–10.6) | 0.57 |
| Diabetic nephropathy, n (%) | 260 (28.8) | 74 (29.5) | 0.88 |
| History of cardiovascular diseases, n (%) | 152 (16.9) | 56 (22.3) | 0.05 |
| Body mass index, kg/m2 | 21.5 (20.0–23.0) | 20.6 (18.5–21.5) | ＜0.001 |
| Kt/V for urea | 1.56 (1.43–1.74) | 1.56 (1.40–1.71) | 0.55 |
| nPCR, g/kg/day | 0.95 (0.85–1.05) | 0.95 (0.84–1.05) | 0.50 |
| Systolic blood pressure, mmHg | 152 (139–166) | 162 (149–174) | ＜0.001 |
| Cardiothoracic ratio, % | 49.2 (45.8–52.4) | 52.1 (48.6–55.7) | ＜0.001 |
| Blood tests | |||
| Hemoglobin, g/dL | 10.6 (10.0–11.2) | 10.5 (9.7–11.2) | 0.12 |
| Serum albumin, g/dL | 4.0 (3.7–4.2) | 3.9 (3.8–4.2) | 0.92 |
| Serum total cholesterol, mg/dL | 158 (137–183) | 152 (131–179) | 0.03 |
| Serum creatinine, mg/dL | 10.2 (8.6–12.2) | 9.8 (8.2–11.9) | 0.05 |
| Serum C-reactive protein, mg/dL | 0.13 (0.05–0.28) | 0.13 (0.04–0.24) | 0.11 |
| Corrected serum calcium, mg/dL | 9.3 (8.8–9.8) | 9.3 (8.9–9.8) | 0.60 |
| Serum phosphate, mg/dL | 4.8 (4.1–5.5) | 4.9 (4.1–5.7) | 0.29 |
| Serum PTH (intact assay), pg/mL | 107 (49–212) | 109 (45–204) | 0.86 |
| Medications | |||
| Use of anti-hypertensives, n (%) | 523 (58.0) | 192 (76.5) | ＜0.001 |
| Use of ESAs, n (%) | 734 (81.4) | 228 (90.8) | ＜0.001 |
| Use of phosphate binders, n (%) | 786 (87.1) | 219 (87.3) | 1.00 |
| Use of VDRAs, n (%) | 641 (71.1) | 180 (71.7) | 0.88 |
Data are shown as median (interquartile range) or number (percentage). ESA, erythropoiesis-stimulating agent; LVH, left ventricular hypertrophy; nPCR, normalized protein catabolic rate; PTH, parathyroid hormone; VDRA, vitamin D receptor activator.
The ECG parameters and the prevalence of abnormal ECG findings are listed in Table 2. The median heart rate was 72 bpm. The prevalence of PR prolongation, QRS prolongation, QTc prolongation, and LVH as per the Sokolow–Lyon voltage criteria was 8.8%, 20.2%, 21.7%, and 21.8%, respectively. Simple correlation coefficients were determined to examine the associations between the ECG parameters (Supplemental Table 7). Heart rate was significantly negatively correlated with the PR interval and QRS duration and positively correlated with the QTc interval. The PR interval was significantly positively correlated with the QRS duration, QTc interval, and sV1＋rV5. The QTc interval was also positively correlated with the QRS duration and sV1＋rV5.
| Parameters | Summary data |
|---|---|
| Heart rate, beat per minute | 72 (64–80) |
| PR interval, msec | 168 (151–191) |
| PR prolongation (≥ 220 msec), n (%) | 101 (8.8) |
| QRS duration, msec | 98 (90–106) |
| QRS prolongation (≥ 110 msec), n (%) | 233 (20.2%) |
| QTc interval, msec | 432 (416–451) |
| QTc prolongation, n (%) (Male ≥ 450 msec, Female ≥ 460 msec) | 250 (21.7) |
| ECG-LVH (SV1+RV5 ≥ 35 mm), n (%) | 251 (21.8) |
The data are shown as the median (interquartile range) or number (percentage).
ECG, electrocardiography; LVH, left ventricular hypertrophy.
| Heart rate | PR interval | QRS duration | QTc interval | |||||
|---|---|---|---|---|---|---|---|---|
| r | P-value | r | P-value | r | P-value | r | P-value | |
| PR interval | -0.128 | ＜0.001 | ||||||
| QRS duration | -0.081 | 0.006 | 0.171 | ＜0.001 | ||||
| QTc interval | 0.161 | ＜0.001 | 0.164 | ＜0.001 | 0.396 | ＜0.001 | ||
| sV1+rV5 | -0.18 | 0.541 | 0.071 | 0.017 | 0.022 | 0.462 | 0.122 | ＜0.001 |
Spearman’s rank correlation coefficient was calculated. A 2-tailed P, 0.05 was considered statistically significant.
During the median follow-up period of 9.0 years (interquartile range 5.3–10.0 years), 518 patients (45%) died of other causes, whereas 76 patients died due to sudden death (14.7% of all deaths). The associations between abnormal ECG findings and the estimated risk of sudden death are summarized in Table 3. In the multivariate-adjusted Cox proportional hazards model, the risk of sudden death was 2.02-fold higher (95% CI 1.05–3.89) at the highest heart rates (Q4) than in the reference quartile (Q2). QTc prolongation (HR 2.10, 95% CI 1.30–3.77) and LVH as per the Sokolow–Lyon voltage criteria (HR 1.91, 95% CI 1.18–3.09) were also determined to be independent predictors of sudden death, even after adjustment for potential confounding factors. The cut-off value of the JTc interval for predicting sudden death was 336 msec, with sensitivity, specificity, and AUC of 0.57, 0.545, and 0.579, respectively (Supplementary Fig.1). The prolonged JTc intervals (≥ 336 msec) were also significantly associated with an increased risk of sudden death (HR 1.82, 95% CI 1.14–2.90). When the ECG parameters were set as continuous variables, the multivariate-adjusted association depicted by the restricted cubic spline curve demonstrated that the HRs for sudden death increased as the baseline heart rate, QTc interval, and RV5/6＋SV1 increased, while neither the PR interval nor the QRS duration showed a significant association with sudden death (Supplementary Figs.2a–2e). We have additionally analyzed the relationship between the ECG parameters and prognosis in the short-term period. As the number of sudden deaths was small in the short-term period, the association between ECG parameters and all-cause mortality during the 4-year follow-up period was examined. Supplemental Table 8 shows the multivariate Cox proportional HRs for all-cause mortality based on the 4-year outcome. The associations between the ECG parameters and all-cause mortality at 4 years were similar at 10 years.
| Model 1 | Model 2 | Model 3 | Model 4 | |||||
|---|---|---|---|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | HR (95% CI) | P-value | |
| Heart rate | ||||||||
| Q1 (≤ 64 bpm), n= 279 | 1.33 (0.67–2.67) | 0.41 | 1.24 (0.62–2.47) | 0.54 | 1.12 (0.56–2.26) | 0.74 | 1.15 (0.57–2.30) | 0.70 |
| Q2 ( 65–72 bpm), n= 279 | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | 1.00 (reference) | ||||
| Q3 (73–80 bpm), n= 283 | 1.17 (0.57–2.39) | 0.67 | 1.29 (0.63–2.64) | 0.49 | 1.21 (0.59–2.49) | 0.50 | 1.15 (0.56–2.36) | 0.71 |
| Q4 (≥ 81 bpm), n= 312 | 1.85 (0.97–3.52) | 0.06 | 2.11 (1.10–4.03) | 0.02 | 1.98 (1.03–3.80) | 0.04 | 2.02 (1.05–3.89) | 0.04 |
| PR prolongation (≥ 220 msec) | 1.20 (0.52–2.77) | 0.67 | 0.85 (0.36–1.98) | 0.70 | 0.74 (0.32–1.76) | 0.50 | 0.73 (0.31–1.73) | 0.48 |
| QRS prolongation (≥ 110 msec) | 1.32 (0.81–2.32) | 1.16 (0.68–1.98) | 0.58 | 1.07 (0.62–1.82) | 0.82 | 0.95 (0.55–1.66) | 0.89 | |
| QTc prolongation | 2.65 (1.66–4.23) | ＜0.001 | 2.29 (1.43–3.68) | 0.006 | 2.06 (1.28–3.33) | 0.003 | 2.10 (1.30–3.77) | 0.01 |
| (Male ≥ 450 msec, Female ≥ 460 msec) | ||||||||
| JTc prolongation (≥ 336 msec) | 1.80 (1.14–2.83) | 0.01 | 1.94 (1.22–3.07) | 0.005 | 1.79 (1.12–2.85) | 0.01 | 1.82 (1.14–2.90) | 0.01 |
| ECG-LVH (SV1+RV5 ≥ 35 mm) | 2.13 (1.33–3.42) | 0.002 | 2.10 (1.31–3.38) | 0.002 | 1.98 (1.23–3.20) | 0.005 | 1.91 (1.18–3.09) | 0.008 |
Model 1, unadjusted model. Model 2, adjusted for age and sex. Model 3, adjusted for the variables in Model 2 plus the presence of diabetic nephropathy, history of cardiovascular events, duration of hemodialysis, and systolic blood pressure. Model 4, adjusted for the covariates in Model 3 plus the serum albumin, total cholesterol, corrected calcium, and phosphate concentrations. A two-tailed P-value of ＜0.05 was considered statistically significant. CI, confidence interval; ECG, electrocardiography; HR, hazard ratio; LVH, left ventricular hypertrophy; Q, quartile of heart rate.
Receiver operating characteristic curve for evaluating the cut-off JTc interval to predict sudden death
Multivariate-adjusted restricted cubic spline plots of hazard ratios for sudden death according to heart rate (a), PR interval (b), QRS duration (c), QT interval (d), and RV5/6＋SV1 €. The solid line represents the hazard ratio, while the gray zone represents the 95% confidence interval. The dotted line corresponds to the normal reference hazard ratio of 1.0. The overall median value was chosen as the reference. The multivariate-adjusted model was adjusted for age, sex, presence of diabetic nephropathy, history of cardiovascular events, duration of hemodialysis, systolic blood pressure, serum albumin, total cholesterol, corrected calcium, and phosphate concentrations.
| HR (95% CI) | P-value | |
|---|---|---|
| Heart rate | ||
| Q1 ( ≤ 64 bpm), n = 279 | 1.00 (reference) | |
| Q2 ( 65–72 bpm), n = 279 | 1.04 (0.66–1.66) | 0.86 |
| Q3 (73-80 bpm), n = 283 | 1.57 (1.03–2.40) | 0.04 |
| Q4 ( ≥ 81 bpm), n = 312 | 1.59 (1.05–2.42) | 0.03 |
| PR prolongation (≥ 220 msec) | 1.44 (0.97-2.14) | 0.07 |
| QRS prolongation (≥ 110 msec) | 1.29 (0.92-1.79) | 0.14 |
| QTc prolongation | 1.82 (1.33-2.50) | 0.002 |
| (Male ≥ 450msec, Female ≥ 460 msec) | ||
| ECG-LVH (SV1+RV5 ≥ 35 mm) | 1.75 (1.27-2.41) | 0.006 |
Covariates included age, sex, the presence of diabetic nephropathy, a history of cardiovascular events, hemodialysis vintage, systolic blood pressure, serum levels of albumin, total cholesterol, corrected calcium, and phosphate. A two- tailed P-value ＜0.05 was considered statistically significant. Abbreviations: CI, confidence interval; ECG, electrocardiogram; HR, hazard ratio; LVH, left ventricular hypertrophy; Q, quartile of heart rate.
We defined the abnormal ECG findings of higher heart rate (≥ 80 bpm), QTc prolongation, and LVH as ECG risk factors for sudden death. The multivariate-adjusted HRs for sudden death according to the number of ECG risk factors are shown in Table 4. The risk of sudden death incrementally increased as the number of ECG risk factors increased.
| Population at Risk | No. of Sudden deaths | HR (95% CI) | P-value | P-value for trend | |
|---|---|---|---|---|---|
| Number of ECG risk factors | |||||
| 0 | 539 | 24 | 1.00 (reference) | − | ＜0.001 |
| 1 | 439 | 28 | 1.57 (0.91–2.73) | 0.11 | |
| 2 | 151 | 19 | 3.15 (1.70–5.83) | ＜0.001 | |
| 3 | 24 | 5 | 6.41 (2.37–17.37) | ＜0.001 |
The ECG risk factors included higher heart rate (≥ 80 bpm), QTc prolongation, and LVH. The multivariate-adjusted model was adjusted for age, sex, presence of diabetic nephropathy, history of cardiovascular events, duration of hemodialysis, systolic blood pressure, and serum albumin, total cholesterol, corrected calcium, and phosphate concentrations. CI, confidence interval; ECG, electrocardiography; HR, hazard ratio; LVH, left ventricular hypertrophy. A two-tailed P value of ＜0.05 was considered statistically significant.
Subgroup analyses stratified by several potential confounding factors were performed to assess the consistency of the association between the baseline ECG parameters and sudden death (Figs.3a–3c). The association between higher heart rate and sudden death was accentuated in patients with longer dialysis duration (P-value for interaction =0.04; Fig.3a). Additionally, the association between QTc prolongation and sudden death was significantly enhanced in patients with a higher serum phosphate concentration (≥ 5.0 mg/dL) (P-value for interaction =0.05; Fig.3b). No significant interactions were observed for any subgroups in relation to LVH on ECG and sudden death (P-value for interaction =0.07–0.95; Fig.3c).
Multivariate-adjusted HRs and 95% CIs for sudden death according to (a) an increase in the baseline heart rate by 10 bpm, (b) QTc prolongation, and (c) LVH according to the Sokolow–Lyon voltage criteria in the subgroups by baseline characteristics. The gray circles and filled rhombi denote the point estimates of the HRs, and the error bars represent the 95% CIs. Adjustment was performed for age, sex, presence of diabetic nephropathy, history of cardiovascular events, duration of hemodialysis, systolic blood pressure, and serum albumin, total cholesterol, corrected calcium, and phosphate concentrations. Variables relevant to the subgroups were excluded from each model. A two-tailed P-value of ＜0.05 was considered statistically significant. CI, confidence interval; HR, hazard ratio; LVH, left ventricular hypertrophy.
In this study, we demonstrated the relationship between baseline ECG parameters and sudden death in patients undergoing hemodialysis. Although there were significant correlations between some ECG parameters, these correlations were deemed not strong. Accordingly, we did not consider the interactions among the ECG parameters. Instead, we determined the association of each ECG parameter with sudden death separately. During the 10-year observation period, sudden death accounted for 14.7% of all deaths in hemodialysis patients. Higher heart rate, QTc prolongation, and LVH, as defined by the Sokolow–Lyon voltage criteria, were significantly associated with an increased risk of sudden death after rigorous adjustment for potential confounding factors using the Cox proportional hazards model. Moreover, the risk of sudden death was noted to significantly increase as the number of ECG risk factors increased. In contrast, neither the PR interval nor the QRS duration was a significant risk factor for sudden death in our cohort. According to previous reports, QRS prolongation is regarded as a marker of cardiac dyssynchrony and is a predictor of cardiac death in patients with congestive heart failure17) and in the general population18). However, in this present study, neither the PR interval nor the QRS duration was a significant risk factor for sudden death. Why the association between the QRS duration and sudden death was not significant in this study is not fully understood. Previous studies have indicated that the QRS duration significantly increases after hemodialysis as a result of changes in body fluid volume and electrolytes19, 20). We speculate that the QRS duration among patients undergoing hemodialysis may not be as accurate at detecting organic and electrophysiological cardiac changes as in nondialysis patients. Our results highlight the usefulness of predialysis resting ECG as a tool to predict sudden death in patients undergoing maintenance hemodialysis. In addition, some ECG parameters predicted prognosis at the 4-year and 10-year follow-ups. In this present study, we suggest that ECG findings predict not only short-term prognosis but also long-term prognosis. ECG parameters are influenced by both fluctuating and reversible factors, such as vital signs, fluid volume, and electrolytes, and by irreversible factors, such as structural changes in the vascular system and myocardium. In this study, the abnormal ECG findings that reflect patient prognosis were not fully concluded. To elucidate these relationships, well-designed prospective studies should collect ECG data, vital signs, and laboratory and imaging data over time and evaluate prognosis using time-averaged or time-varying models.
In this study, we demonstrated the association between higher heart rate and the increased risk of sudden death. Elevated heart rate can occur with sympathetic nervous system activation, worsening myocardial ischemia, increased systemic inflammation, and endothelial dysfunction, which can all lead to sudden death21). Several studies have shown that a higher resting heart rate is an independent risk factor for sudden death in the general population22, 23). Moreover, a higher heart rate has been associated with an increased risk of all-cause mortality in hemodialysis patients24). However, few studies have focused on the association between heart rate and sudden death in hemodialysis patients. To the best of our knowledge, this study is the first to show that an elevated prehemodialysis heart rate measured by ECG is associated with a higher risk of sudden death in hemodialysis patients. Unfortunately, we did not obtain detailed information about patients’ medications, which could affect their heart rate, such as β-blockers. Therefore, further studies are necessary to confirm our observations.
In this study, 20% of all patients showed QTc interval prolongation. Notably, the prevalence of QTc interval prolongation was lower than in previous studies of patients undergoing hemodialysis25). The QTc interval reflects the duration of ventricular depolarization and repolarization and is influenced by various factors, such as electrolyte shifts, drugs, myocardial ischemia, and structural heart disease. QTc interval prolongation may induce an extended vulnerable period and increase the risk of ventricular tachyarrhythmias, such as Torsade de pointes, resulting in sudden death. While several previous studies have shown that prolonged QTc intervals are associated with sudden death in the general population26, 27), studies focusing on the relationship between QTc interval prolongation and the risk of sudden death remain to be limited. A few studies have reported an association between QTc interval prolongation and an increased risk of mortality and sudden death in patients undergoing hemodialysis28, 29). However, these studies had small sample sizes and relatively short follow-up periods. Our study had a larger sample size and a longer 10-year follow-up period, which may provide robust evidence for the association between QTc interval prolongation and the increased risk of sudden death in hemodialysis patients. In this study, JTc interval prolongation was also associated with sudden death. Although previous studies have shown the associations of the JTc interval with all-cause mortality30) and cardiovascular disease31), the relationship between the JTc interval and sudden death has not been reported previously. We thus suggest that the JTc interval might be a useful predictor of sudden death among patients undergoing hemodialysis. As a trial, we determined the cut-off value of JTc for predicting sudden death (336 msec) by performing ROC curve analysis in this study. However, the cut-off value for the JTc interval prolongation has not been defined internationally. Hence, introduction of the JTc interval prolongation may be limited at this point. Therefore, we think that further studies are needed to determine the optimal range of JTc interval.
LVH is one of the most common complications in patients with end-stage kidney disease. LVH causes sudden death via subendocardial ischemia and cardiac fibrosis32). In this present study, we demonstrated that LVH was prevalent in 20% of patients in our hemodialysis cohort. The prevalence of LVH in our cohort was higher than in previous reports on dialysis patients33) and nondialysis patients34). LVH measured by echocardiography is a strong predictor of sudden death35). Although we did not confirm LVH by echocardiography in this study, LVH on ECG is noted to be closely correlated with left ventricular mass index determined by echocardiography36). Furthermore, LVH on ECG is associated with sudden death among hemodialysis patients with diabetes mellitus33). Importantly, the study by Krane et al. was limited to patients with diabetes mellitus undergoing maintenance hemodialysis, whereas our study demonstrated that LVH on ECG is a predictor of sudden death in both diabetic and nondiabetic patients undergoing hemodialysis.
Our subgroup analysis showed that the effects of heart rate on sudden death were enhanced in patients with longer dialysis duration (≥ 5 years). Patients with longer dialysis duration might have a poor nutritional state, reduced cardiac function, advanced atherosclerosis, arteriosclerosis, and LVH, all of which could trigger cardiovascular events. These comorbidities could intensify the impact of higher heart rate on sudden death.
The subgroup analysis also showed that the risk of sudden death in patients with QTc prolongation was significantly higher in patients with an elevated baseline serum phosphate concentration (≥ 5.0 mg/dL). We have previously reported that a higher serum phosphate concentration is associated with an increased risk of sudden death37). Hyperphosphatemia may be directly linked to sudden death and indirectly via vascular calcification, endothelial dysfunction, and LVH by increasing serum fibroblast growth factor 23 levels38, 39). Although plausible biological explanations for this interaction are limited, we speculate that the combination of QTc prolongation and high serum phosphate concentration synergistically increases the incidence of sudden death.
The strengths of our study are the relatively large sample size and the relatively long observation period compared with previous studies. However, several limitations of this study should be noted. First, ECG was recorded only at baseline and predialysis. Therefore, we could not assess ECG changes during the hemodialysis sessions and over the observation period. Second, autopsy was not performed in all patients; therefore, the exact cause of sudden death in each patient was not determined. Third, because we evaluated only ECG parameters measured automatically by the ECG machines in this study, other detailed ECG information, which could influence the interpretation of each ECG parameter, such as ST–T changes and right or left bundle branch block, were not considered. Fourth, the timing of blood sampling and ECG did not match exactly because we retrospectively collected the ECG data recorded within 1 year of study enrollment after the 10-year follow-up period. Therefore, the laboratory data did not necessarily reflect the ECG findings. Fifth, we did not have information on some parameters; serum potassium and magnesium concentrations, which could be a strong predictor of the QT interval; cardiac function measured by echocardiography (e.g., ejection fraction and LVH), which could be associated with some ECG parameters; and medical treatment (e.g., β-blockers and antiarrhythmic drugs), which could influence heart rate, were missing. Therefore, they could be considered as residual confounding factors. Sixth, approximately two-thirds of the participants (2,445 of the 3,598 potentially eligible participants) were excluded. Significant differences in some baseline characteristics were found between the included patients and the excluded patients, which may limit the generalizability of the results to all hemodialysis patients. Finally, the ECG data were obtained retrospectively from multiple centers; therefore, the ECG devices and measurement methods were not standardized. Moreover, each parameter was automatically measured with each device, which might have led to measurement error and misclassification bias. Despite these limitations, we believe that our study shows the prognostic value of periodic ECG recording and provides useful information for medical practitioners to stratify hemodialysis patients who are at an increased risk of sudden death.
In this study, we showed that ECG parameters, including higher heart rate, QTc prolongation, and LVH as per the Sokolow–Lyon voltage criteria, are independently associated with an increased risk of sudden death in hemodialysis patients. Our study suggests that ECG recorded in the resting state before hemodialysis may be a valuable risk stratification tool to predict sudden death in patients undergoing maintenance hemodialysis. Further well-designed, large-scale, interventional trials are required to confirm whether maintaining each ECG parameter within an optimal range could reduce the risk of sudden death in hemodialysis patients.
The authors would like to express their appreciation to the investigators at the participating institutions. We also thank the participants from the Q-Cohort Study and members of the Society for the Study of Kidney Disease. The following personnel (institutions) participated in the study: Takashi Ando (Hakozaki Park Internal Medicine Clinic), Takashi Ariyoshi (Ariyoshi Clinic), Koichiro Goto (Goto Clinic), Fumitada Hattori (Nagao Hospital), Harumichi Higashi (St. Mary’s Hospital), Tadashi Hirano (Hakujyuji Hospital), Kei Hori (Munakata Medical Association Hospital), Takashi Inenaga (Ekisaikai Moji Hospital), Hidetoshi Kanai (Kokura Memorial Hospital), Shigemi Kiyama (Kiyama Naika), Tetsuo Komota (Komota Clinic), Hiromasa Kuma (Kuma Clinic), Toshiro Maeda (Kozenkai-Maeda Hospital), Junichi Makino (Makino Clinic), Dai Matsuo (Hirao Clinic), Chiaki Miishima (Miishima Naika Clinic), Koji Mitsuiki (Japanese Red Cross Fukuoka Hospital), Kenichi Motomura (Motomura Naika Clinic), Sadatoshi Nakamura and Hidetoshi Nakamura (Kokura Daiichi Hospital), Koichi Nakashima (Ohashi Internal Circulatory Clinic), Nobumitsu Okita (Shiroishi Kyoritsu Hospital), Shinichiro Osato (Osato Jin Clinic), Sakura Sakamoto (Fujiyamato Spa Hospital), Keiko Shigematsu (Shigematsu Clinic), Kazumasa Shimamatsu (Shimamatsu Naika Iin), Yoshito Shogakiuchi (Shin-Ai Clinic), Hiroaki Takamura (Hara Hospital), Kazuhito Takeda (Iizuka Hospital), Asuka Terai (Chidoribashi Hospital), Hideyoshi Tanaka (Mojiko-Jin Clinic), Suguru Tomooka (Hakozaki Park Internal Medicine Clinic), Jiro Toyonaga (Fukuoka Renal Clinic), Hiroshi Tsuruta (Steel Memorial Yawata Hospital), Ryutaro Yamaguchi (Shiseikai Hospital), Taihei Yanagida (Saiseikai Yahata General Hospital), Tetsuro Yanase (Yanase Internal Medicine Clinic), Tetsuhiko Yoshida (Hamanomachi Hospital), Takahiro Yoshimitsu (Gofukumachi Kidney Clinic, Harasanshin Hospital), and Koji Yoshitomi (Yoshitomi Medical Clinic).
We thank Emily Woodhouse, PhD, and Leah Cannon, PhD, from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript.
This study was supported by the Kidney Foundation (H19 JKFB 07-13, H20 JKFB 08-8, H23 JKFB 11-11) and the Japan Dialysis Outcome Research Foundation (H19-076-02, H20-003). The funders of this study had no role in study design; collection, analysis, and interpretation of data; writing the report; and the decision to submit the report for publication.
The individual contributions of each co-author were as follows. Hiroto Hiyamuta and Shunsuke Yamada contributed to the study design, statistical analysis, interpretation of data, and drafting of the manuscript. Masatomo Taniguchi and Kosuke Masutani contributed to the acquisition of data and critical revision of the manuscript. Toshiaki Nakano contributed to funding, acquisition of data, and critical revision of the manuscript. Kazuhiko Tsuruya and Takanari Kitazono contributed to the critical revision of the manuscript and supervision of the study. All authors provided critical reviews of the draft and approved the final version.
The authors declare that they have no other relevant financial interests.
