Rationale and design for the Predictors of Arrhythmic and Cardiovascular Risk in End Stage Renal Disease (PACE) study

We conducted ECG recording at baseline and annual follow-up visits. We defined specific ECG parameters in Table 2. A standard 12-lead ECG was recorded at rest in supine position using the Marquette MAC 5500 HD ECG system (GE Medical Systems, Milwaukee, WI). High resolution (1000 Hz). Orthogonal Frank XYZ ECG was recorded using the Norav 1200 M PC ECG machine (Norav Medical Ltd, Thornhill, ON, Canada) at rest in the supine position for at least 5 minutes. Only sinus beats were included in the analysis. Subjects with serious cardiac arrhythmias defined as ventricular tachycardia, ventricular fibrillation, and atrial fibrillation/flutter were captured, if detected, based on standard criteria [9]. Atrial fibrillation/flutter, as well as premature atrial and ventricular contractions with the first subsequent beat were excluded from subsequent analyses. Fiducial points were automatically detected on each ECG lead and verified by investigators as previously described [66]. Bazett (QT_B) [67], Fridericia (QT_F) [68], and individualized (QT_i) [69] approaches for QT correction were applied. Both time domain and frequency domain HRV analyses were performed as recommended by the European Society of Cardiology/North American Society of Pacing and Electrophysiology (ESC/NASPE) Task Force, as well as the American College of Cardiology (ACC)/ American Heart Association (AHA) guidelines for analysis of short-term recordings [9]. Robust automated template matching techniques were used for QT variability analyses [70,71].

Additionally, SAECG was obtained with a MAC 5500 HD ECG system (GE Medical Systems, Milwaukee, WI), with bandpass filtering of 40–250 Hz, and averaging of 350 QRS complexes. The following SAECG characteristics of the filtered QRS were evaluated [72,73]: (1) total duration (fQRSd), (2) duration of the low-amplitude signals (<40 mV) in the terminal portion (LAS40), and (3) root-mean-square voltage of the last 40 ms (RMS40). VLPs were considered positive when ≥2 of the following criteria were fulfilled: (1) fQRSd >114 ms, (2) LAS40 > 38 ms, and (3) RMS40 < 20 μV.

All ECG measures are analyzed as continuous variables or clinical categories based on standardized methods [9,55], and are studied as potential effect modifiers [74].

All subclinical CVD measures used standardized protocols similar to the CRIC Study (Table 2) [63,64] with corresponding quality control procedures. We conducted the CT at baseline on each participant to determine coronary artery calcium and arterial stenoses. Agatston coronary calcification scores, volume, volumetric scores, and mass for the four main coronary arteries and cardiac valves were calculated. Aortic or mitral valvular calcification was also defined as present or not. Coronary artery stenosis was defined as significant narrowing (50% or more diameter reduction) of the lumen of the four main coronary branches: left main, left anterior descending, left circum-flex, and right coronary artery, including side branches according to the AHA classification [75]. The effective radiation dose estimation for the entire study that includes CT angiography and calcium measurement was approximately 5–7 mSv (~0.7 rem). Prior to the study, a beta-blocker, metoprolol, was given orally to bring the heart rate to less than 65 beats per minute. Based on protocol, trained cardiologists and radiologists independently read all CT images to assess for both cardiac and non-cardiac pathologies and submitted a separate report of these findings within a week [76].

Echocardiograms were conducted by three trained echocardiographers at baseline on each participant to determine LV stroke volume, LV ejection fraction, and LV mass index with four chamber views and standard calculations for LV mass. The M-mode by the parasternal short axis view was used to estimate LV mass, as the long axis view can result in improper alignment and overestimate LV dimensions and mass [77]. Additionally, we collected pulmonary artery pressures based on flow at the tricuspid valve. This was not always possible based on body habitus and lack of regurgitation evident at the valve. All cardiovascular studies were centrally read at the Johns Hopkins Cardiovascular Laboratory.

Aortic pulse wave velocity (PWV) measures were performed supine after at least 5 minutes of rest using the right carotid and right femoral arteries by four trained research staff [78]. The operator captured 10 seconds of stable carotid waveform and repeated the sequence using the femoral artery. After the second waveform was captured, the computer generated an estimate of aortic PWV with a standard deviation. If the standard deviation was more than 15% of the PWV value, the study was repeated. Pulse wave analysis was also measured using tonometry of the right radial artery (and left if arteriovenous fistula was present). Quality control was assessed in real-time using quality indices and operator index for each waveform generated. Additional internal quality control included review of study procedures by operator every 6 months to ensure adherence to the protocol.

Ankle brachial index (ABI) was performed with the patient resting for 5–10 minutes in a supine position and assessed by measuring blood pressures with a Doppler probe bilaterally in the brachial, dorsalis pedis, and posterior tibial arteries. The right and left ankle–brachial index values were determined by dividing the higher ankle pressure in each leg by the higher arm pressure.

We also obtained discharge summaries for all hospitalizations as reported by participants or dialysis unit. In addition, surveillance of hospitals commonly frequented by participants was conducted to minimize missed emergency room visits, hospitalizations, or deaths. Also, linkage to the United States Renal Data Systems (USRDS) in the future will allow for additional data capture on healthcare utilization.

We define composite outcomes of cardiovascular disease (Table 3). Atherosclerotic disease is classified by standardized definitions for myocardial infarction, coronary revascularization procedures, stroke, carotid endarterectomy, congestive heart failure, and peripheral vascular disease by amputation or peripheral surgical/ percutaneous. Clinical or symptomatic arrhythmic events include treated atrial fibrillation/atrial flutter, heart block or bradycardias, ventricular fibrillation or tachycardia, and additional procedures for a pacemaker or defibrillator or electrocardioversion or ablation. Composite of arrhythmic disease is: (1) atrial arrhythmias including atrial fibrillation by annual ECGs and new arrhythmic events such as treated atrial fibrillation by electrocardioversion or ablation or rate control by pharmacological intervention, and death related to atrial fibrillation [79,80]; and (2) ventricular-related arrhythmias including ventricular tachycardia by annual ECGs, SCD, hospitalized arrhythmias, and placement of implantable defibrillators [81].

From our regular contact with the dialysis unit, we have been informed of a participant’s death. Once notified of a participant’s death, we obtained the CMS death notification form (CMS-2746). We also interviewed the next of kin to determine when the participant was last seen prior to death or the last hospitalization and whether any symptoms preceded the event. All records from the hospitalization or emergency room visit were also obtained. To ensure that death ascertainment is complete, we searched the National Death Index annually for persons lost to follow-up [82].

We used similar criteria for death classification adapted from the HEMO and CRIC studies [83]. SCD was defined as a sudden pulseless condition (collapse or syncope) presumed to be due to an arrhythmia occurring out of the hospital or in the emergency room in an otherwise stable individual. In case of an unwitnessed event, there was an evidence that the patient was seen in a stable condition within the 24 hours preceding the event (or since the last dialysis session). All events that occurred during a hospitalization, or in nursing home, or hospice were not classified as SCD. Deaths attributed to coronary artery disease as similarly defined in the Atherosclerosis Risk in Communities (ARIC) Study [84] if the patients (1) have had a definite myocardial infarction within 4 weeks of death, or (2) have had chest pain within 72 hours of death in cases of out-of-hospital death or cardiac pain in cases of in-hospital death, or (3) history of chronic ischemic heart disease such as myocardial infarction, coronary insufficiency, or angina pectoris, or (4) the underlying cause of death in the death certificate included ICD-10 code I20, I21, I22, I23, I24, I25, I46, I51.6, I51.9, R99, J96, if there was no evidence of a non-atherosclerotic or non-cardiac atherosclerotic process that was the probable cause of death.

We had an independent Data Safety Monitoring Committee with three physicians in nephrology and cardiology who reviewed any adverse events related to study procedures twice a year. The Committee reviewed events that occurred during the study visits or from the study’s cardiac imaging procedures. Clinical readings of the echocardiogram, CT angiography, and CT chest over-reads were required even though they were obtained for clinical research, as the discovery of incidental findings may have necessitated the timely reporting of critical results to the primary nephrologists, dialysis facilities, primary care physicians, and participants [76]. Depending on the findings, we had instituted a protocol to report urgent findings that required intervention to the nephrologist and participant.