There were 118 patients with Fontan palliation identified, and a total of fifty patients, with a median age of 27 years (IQR 23,31), 29 (58%) male, were enrolled in this study (Fig. 1). By Fontan type, 33 (66%) patients had lateral tunnel Fontan palliation, and 32 (64%) had a single left ventricle (Table 2). Most patients were NYHA Class I (17, 34%) or II (26, 52%); 45 (92%) patients were ACHD physiologic stage C. The majority of patients had normal or mildly reduced systemic ventricle function (45, 94%) with normal or mild atrial-ventricular valve (AVV) regurgitation (45, 94%). Over half (28, 56%) of the study population had a history of any arrhythmia, either sustained or non-sustained, and 6% of patients had a previous atrial arrhythmia requiring cardioversion. Most patients had a Holter within one year of enrollment (38, 76%), and 16% of those patients had non-sustained SVT detected on recent Holter.
Table 2
Demographics and clinical characteristics. AP = atrio-pulmonary; ICD = implantable cardioverter defibrillator.
Characteristic | N = 50 |
Age at Enrollment | 27 (23, 31) |
Gender | |
Male | 29 (58%) |
Female | 21 (42%) |
Underlying cardiac diagnosis | |
Single RV | 13 (26%) |
Single LV | 32 (64%) |
Mixed | 5 (10%) |
Type of Fontan Palliation | |
Lateral Tunnel | 33 (66%) |
Extracardiac | 16 (32%) |
AP | 1 (2.0%) |
Systemic ventricle Echo function: | |
Normal/mildly reduced | 45 (94%) |
Moderate | 3 (6.3%) |
Severely depressed | 0 (0%) |
AVV Regurgitation | |
Normal/Mild | 45 (94%) |
Moderate | 3 (6.3%) |
Severely depressed | 0 (0%) |
Unknown NYHA Function Class Class I Class II Class III Class IV | 2 17 (34%) 26 (52%) 7 (14%) 0 (0%) |
Pulmonary Hypertension | 1 (2.0%) |
Creatinine above cutoff for Sex | 5 (11%) |
Pulse Ox (%) | 91.0 (89.0, 93.0) |
Unknown | 3 |
History of any arrhythmia, sustained or non-sustained | 28 (56%) |
History of Sustained Atrial Flutter | 3 (6.0%) |
Symptoms at the time of enrollment? | 11 (22%) |
Pacemaker or ICD | 13 (26%) |
Holter within 1 year of enrollment | 38 (76%) |
Values are presented as median (IQR) or n (%) |
Enrollment and Patient Driven KM Tracings
During the study period, there were 50 enrollment tracings and 510 follow up transmissions received from 34 (68%) patients, 449 of which were feasible for analysis, including 422 asymptomatic transmissions and 22 symptomatic transmissions. There was a median of 7.5 transmissions sent per patient (Fig. 2). Inter-rater reliability of the KM-EP interpretations had a very strong agreement for follow-up tracings (98% agreement, Cohen Κ = 0.89).
Enrollment KM-auto compared to enrollment EKG
Approximately two-thirds of the enrollment KM tracings 34/50 (68%) were most consistent with lead I of the 12 lead EKG. The QRS and T wave concordance analysis was consistent with the EKG in 96% of enrollment tracings (K 0.891 (0.74,1.0) and 0.9 (0.76,1.0), respectively). Notably, the P wave axis was concordant with the EKG in only 58% of enrollment tracings (Cohen K 0.381 (0.2,0.54)).
Seventeen enrollment KM-auto interpretations (34%) were classified as uninterpretable; and the corresponding enrollment EKGs for these uninterpretable KM-auto enrollment transmissions demonstrated sinus rhythm, ectopic atrial rhythm, sinus rhythm with premature atrial contractions, and atrial flutter in one case. A single KM-auto enrollment interpretation as possible AF was determined to be an atrial-ventricular paced rhythm based upon comparison to the 12 lead EKG. There was 64% percent agreement (Cohens K 0.033 (-0.03,0.097)) between KM-auto and the 12 lead EKG. Sensitivity and specificity of the KM-auto interpretation to detect a normal transmission compared to enrollment 12 lead EKG were 65% and 100%, respectively (Table 3).
Table 3
KM analysis demonstrating the sensitivity, specificity, positive predictive value (PPV), and negative predictive valve (NPV) of KM-auto normal interpretations. KM-Auto = automated KM reads; KM-EP = electrophysiologist KM interpretation.
| Agreement & Concordance Metrics | Diagnostic Metrics for Normal Transmission |
Follow-up Period | Comparison | Percent Agreement | Cohen’s K and 95% CI | Sensitivity | Specificity | PPV | NPV |
Enrollment | EKG vs. KM-EP | 0.90 | -0.03 (-0.07, 0.02) | 0.92 | 0.00 | 0.98 | 0.00 |
EKG vs. KM-Auto | 0.64 | 0.03 (-0.03, 0.01) | 0.65 | 1.00 | 1.00 | 0.06 |
KM-Auto vs KM-EP | 0.68 | 0.19 (0.02, 0.37) | 0.70 | 1.00 | 1.00 | 0.22 |
Follow-up | KM-auto vs. KM-EP | All | 0.72 | 0.13 (0.08, 0.17) | 0.75 | 0.96 | 1.00 | 0.18 |
Symptomatic | 0.50 | 0.20 (0.01, 0.38) | 0.59 | 1.00 | 1.00 | 0.42 |
Asymptomatic | 0.73 | 0.11 (0.06, 0.16) | 0.76 | 0.95 | 1.00 | 0.16 |
Enrollment KM-EP compared to enrollment EKG
There were 46 enrollment KM-EP interpretations that were categorized as normal. Based on EKG, 18 of these 46 were classified as normal sinus rhythm, 18 as normal sinus rhythm with conduction delay, 4 as sinus bradycardia, 4 as ectopic atrial rhythm, 1 as junctional rhythm, and 1 as atrial flutter. Percent agreement between KM-EP and EKG reads was 90% (Cohen’s K -0.03 (-0.07,0.02)). The four KM-EP interpretations not categorized as normal were classified as possible AT in two of the cases or uninterpretable in the other two cases. The EKG for these 4 demonstrated normal sinus rhythm and ectopic atrial rhythm. Sensitivity of the KM-EP interpretations in detecting a normal rhythm was 92.8% (Table 3).
Enrollment KM-EP compared to KM-auto
The number of normal KM enrollment interpretations were higher in the KM-EP compared to the KM-auto reads 46/50 vs 32/50. There were two KM recordings that were uninterpretable by KM-EP evaluation, compared to 17 KM-auto interpretations. Percent agreement between KM-auto and KM-EP reads was only 68% (Cohen’s K 0.194 (0.018,0.37)).
Additionally, enrollment transmissions were evaluated based on those which were concordant and discordant between KM-auto and KM-EP interpretation. History of any sustained or nonsustained arrhythmia was associated with discordant KM-auto and KM-EP interpretations (Table 4). No other clinical factors were found to be significantly associated with discordant interpretations.
Table 4
Demographics and clinical characteristics comparing initial enrollment transmissions that had KM-auto and KM-EP agreement versus disagreement.
| Initial Transmission |
Characteristic | Agreement, N = 34 | Disagreement, N = 16 | p-value |
Age at Enrollment | 28 (23, 30) | 26 (22, 37) | 0.803 |
Gender | | | 0.863 |
Male | 20 (59%) | 9 (56%) | |
Female | 14 (41%) | 7 (44%) | |
Underlying cardiac diagnosis | | | 0.653 |
Single RV | 8 (24%) | 5 (31%) | |
Single LV | 23 (68%) | 9 (56%) | |
Mixed | 3 (8.8%) | 2 (13%) | |
Type of Fontan Palliation | | | 0.303 |
AP | 1 (2.9%) | 0 (0%) | |
Extracardiac | 13 (38%) | 3 (19%) | |
Lateral Tunnel | 20 (59%) | 13 (81%) | |
NYHA Function Class | | | 0.341 |
Class I | 13 (38%) | 4 (25%) | |
Class II | 18 (53%) | 8 (50%) | |
Class III | 3 (8.8%) | 4 (25%) | |
Class IV | 0 (0%) | 0 (0%) | |
ACHD Physiologic Class | | | 0.463 |
A | 1 (2.9%) | 0 (0%) | |
B | 1 (2.9%) | 2 (13%) | |
C | 32 (94%) | 13 (87%) | |
D | 0 (0%) | 0 (0%) | |
Unknown | 0 | 1 | |
Systemic ventricle Echo function: | | | 0.227 |
Normal/mildly reduced | 32 (97%) | 13 (87%) | |
Moderate | 1 (3.0%) | 2 (13%) | |
Severely depressed | 0 (0%) | 0 (0%) | |
Unknown | 1 | 1 | |
AVV Regurgitation | | | > 0.999 |
Normal/Mild | 31 (94%) | 14 (93%) | |
Moderate | 2 (6.1%) | 1 (6.7%) | |
Severely depressed | 0 (0%) | 0 (0%) | |
Unknown | 1 | 1 | |
Pulse Ox (%) | 91.5 (88.8, 93.3) | 90.0 (89.5, 91.0) | 0.396 |
Unknown | 2 | 1 | |
History of arrhythmia (at time of enrollment or prior to enrollment) | 15 (44%) | 13 (81%) | 0.014 |
Symptoms at the time of enrollment? | 8 (24%) | 3 (19%) | > 0.999 |
Pacemaker or ICD | 6 (18%) | 7 (44%) | 0.082 |
Match ECG Lead | | | 0.520 |
I | 25 (74%) | 9 (56%) | |
P-wave matched? | 22 (65%) | 7 (44%) | 0.161 |
Values are presented as median (IQR) or n (%) |
Follow up transmission KM-auto compared to KM-EP
KM-auto interpretations of the asymptomatic transmissions included 306/422 normal, 21/422 possible AF, 27/422 tachycardia, and 68/422 uninterpretable. During the study, 22 symptomatic transmissions were submitted by 11 patients, and the KM-auto reads were 10/22 normal, 3/22 possible AF, 2/22 tachycardia, and 7/22 uninterpretable. The sensitivity and specificity for KM-auto to detect a normal transmission compared to KM-EP interpretations of all follow up transmissions was 75% and 96%, respectively (Table 3). The highest agreement of the KM-auto to KM-EP interpretation was for asymptomatic transmissions (= 73% (Cohen K = 0.112). Of note, one symptomatic KM submission had a KM-auto interpretation of normal but was noted by a clinical team to be a higher heart rate than their normal baseline, and had atrial flutter confirmed on a follow up 12 lead EKG (Fig. 3).
There were 16 (34%) enrolled patients who did not submit any follow up transmissions. Medical records were reviewed for these patients and confirmed no known atrial arrhythmias occurred during the study period.