Portable Device Improves the Detection of Atrial Fibrillation After Ablation
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
CLINICAL STUDY
Portable Device Improves the Detection of Atrial Fibrillation After
Ablation
Songqun Huang,1,* MD, Teng Zhao,1,* MD, Chao Liu,1,* MD, Aihong Qin,1 MD, Shaohua Dong,1 MD,
Binhang Yuan,2 MD, Wenhui Xing,3 MD, Zhifu Guo,1 MD, Xinmiao Huang,1 MD,
Yongmei Cha,4 MD and Jiang Cao,1 MD
Summary
Asymptomatic recurrences of atrial fibrillation (AF) have been found to be common after ablation.
A randomized controlled trial of AF screening using a handheld single-lead ECG monitor (BigThumbⓇ) or
a traditional follow-up strategy was conducted in patients with non-valvular AF after catheter ablation. Consecu-
tive patients were randomized to either BigThumb Group (BT Group) or Traditional Follow-up Group (TF
Group). The ECGs collected via BigThumb were compared using the automated AF detection algorithm, artifi-
cial intelligence (AI) algorithm, and cardiologists’ manual review. Subsequent changes in adherence to oral anti-
coagulation of patients were also recorded. In this study, we examined 218 patients (109 in each group). After a
follow-up of 345.4 ± 60.2 days, AF-free survival rate was 64.2% in BT Group and 78.9% in TF Group (P =
0.0163), with more adherence to oral anticoagulation in BT Group (P = 0.0052). The participants in the BT
Group recorded 26133 ECGs, among which 3299 (12.6%) were diagnosed as AF by cardiologists’ manual re-
view. The sensitivity and specificity of the AI algorithm were 94.4% and 98.5% respectively, which are signifi-
cantly higher than the automated AF detection algorithm (90.7% and 96.2%).
As per our findings, it was determined that follow-up after AF ablation using BigThumb leads to a more
frequent detection of AF recurrence and more adherence to oral anticoagulation. AI algorithm improves the ac-
curacy of ECG diagnosis and has the potential to reduce the manual review.
(Int Heart J 2021; 62: 786-791)
Key words: Rhythm monitoring, Artificial intelligence algorithm, BigThumb, Catheter ablation, Follow-up
strategy
trial fibrillation (AF) has been identified as one Holter monitoring, may overestimate the effectiveness of
A of the most common arrhythmias, affecting an es-
timated 43.6 million patients worldwide.1) It is
considered a serious public health problem because of its
increasing incidence and prevalence in the aging popula-
catheter ablation.4) Meanwhile, the portable devices used
to monitor AF and artificial intelligence (AI) algorithm
applied have been found to have the potential to improve
accuracy in rhythm monitoring and diagnosis.5,6)
tion and its association with elevated risks of cardiovascu- We therefore undertook a randomized controlled trial
lar events and death.2) Catheter ablation is increasingly with the portable device versus traditional follow-up strat-
common in AF treatment because of its ability to reduce egy in a post-ablation population in order to evaluate the
symptoms. However, the number of asymptomatic recur- feasibility and veracity of the self-applied ECG monitor-
rence of AF after ablation has been significant; thus, a fre- ing device in detecting AF.
quent monitoring strategy is warranted for early detection
of AF episode and subsequent change of medical treat-
Methods
ment.3)
Asymptomatic AF is difficult to diagnose based on a This study was designed as a randomized controlled
short electrocardiography (ECG) recording, especially single-center study.
when the episode is paroxysmal. It has been demonstrated Patients: Patients aged >18 years with symptomatic AF
that traditional monitoring methods, including ECG and refractory to at least one class I or class III antiarrhythmic
From the 1Department of Cardiovasology, Changhai Hospital, Second Military Medical University, Shanghai, China, 2Department of Computer Science,
William Marsh Rice University, Houston, USA, 3Shanghai Yueguang Medical Technologies Ltd, Shanghai, China and 4Division of Cardiovascular Diseases,
Mayo Clinic, Rochester, USA.
*These authors contributed equally to this work.
This study was supported by the National Natural Science Foundation of China 81970278.
Address for correspondence: Jiang Cao, MD, Department of Cardiovasology, Changhai Hospital, Second Military Medical University, 168 Changhai Road,
Shanghai, 200433 China. E-mail: chrfca@163.com
Received for publication January 31, 2021. Revised and accepted March 31, 2021.
Released in advance online on J-STAGE July 17, 2021.
doi: 10.1536/ihj.21-067
All rights reserved by the International Heart Journal Association.
786
Int Heart J July 2021 PORTABLE DEVICE IMPROVES ATRIAL FIBRILLATION DETECTION 787 Figure 1. A: The BigThumb system. With two thumbs of the users on the device, a bECG will be collected. B: Software interface of the BigThumb. Users are asked if there is a symptom when bECG is collected. The green means low, and the red means high in the color bar. C: A bECG collected by a patient with atrial fibrillation. D: Summary of bECG in 1-month duration. The red points show the bECG with atrial fibrilla- tion, and the green points show the sinus rhythm. bECG, BigThumb ECG. drug and referred for a first catheter ablation procedure at ministration (NMPA)-approved smartphone handheld tool Changhai Hospital between March 2019 and October that permits easy and rapid collection of a “one-lead” 2019 were enrolled in this study. All patients provided ECG, which we define as BigThumb ECG (bECG) with written informed consent. Consecutive 218 patients re- two thumbs of users on the device (Figure 1). In addition, ceived randomization after ablation; they were then ran- the bECG will also be sent to doctors automatically. Users domly assigned to either BigThumb Group (BT Group) or can store and review bECGs on their cellphones. AI algo- Traditional Follow-up Group (TF Group) by means of rithm was applied in the new version of BigThumb (Sup- random number table. The baseline clinical features were plemental Text). It may improve the accuracy of automatic noted for each patient and compared for different follow- diagnosis. The feasibility of the BigThumb applied in the up strategies. This study was conducted in accordance follow-up of patients after ablation was then characterized. with the Declaration of Helsinki and was approved by the Follow-up strategies: AF recurrence is defined as Holter institutional review board of Shanghai Changhai Hospital, ECG tracing of more than 30 seconds showing heart Second Military Medical University. rhythm with no discernible repeating P-waves and irregu- Ablation procedures: For all cases, standard femoral ve- lar RR intervals. Patients in BT Group were provided with nous access was achieved using Seldinger technique with and trained to use a BigThumb ECG monitor; they were a 6-French and a 7-French sheath in the left femoral vein instructed to take at least three bECGs every day, or more and two 8.5-French Swartz sheaths in the right femoral if symptomatic. The patients should mark symptoms when vein. Coronary sinus catheter and right ventricle catheter bECGs were collected. AF was detected by both an auto- were advanced through the left femoral vein. Transseptal mated AF detection algorithm and an AI algorithm. Two access was achieved with fluoroscopic guidance. A Tacti- cardiologists will confirm the diagnosis separately, and a Cath Quartz catheter (St. Jude Medical, St. Paul, MN, third cardiologist who is more qualified will make the fi- USA) was used for radiofrequency catheter ablation in nal decision in case of disagreement. each procedure. The basic ablation strategy is pulmonary Meanwhile, AF recurrence of patients in TF Group vein isolation, and the specific method for each case was was determined using Holter monitors at 3, 6, and 12 left to the discretion of the primary operator. months after ablation and ECGs if there are symptoms. The BigThumb heart monitor: The BigThumb (Shang- Monitoring data and patient management strategies were hai Yueguang Medical Technologies Inc., Shanghai, then collected and analyzed. A blanking period was de- China) heart monitor is a National Medical Products Ad- fined as 3 months after ablation. Oral anticoagulation was
Int Heart J
788 HUANG, ET AL July 2021
Table I. Characteristics Between Patients in BT and TF Groups
BT Group TF Group
Variables P-value
(n = 109) (n = 109)
Male (%) 84 (77.1%) 70 (64.2%) 0.296
Age (years) 62.3 ± 8.5 64.1 ± 10.3 0.165
Paroxysmal AF (%) 65 (59.6%) 76 (69.7%) 0.119
BMI (Kg/m2) 24.9 ± 2.7 24.5 ± 2.8 0.347
CHA2DS2-VASc 1.7 ± 1.5 2.0 ± 1.4 0.192
HAS-BLED 1.2 ± 1.0 1.3 ± 1.0 0.889
Scr (umol/L) 77.4 ± 15.3 79.7 ± 22.3 0.375
GFR (mL/minute) 88.4 ± 15.3 84.3 ± 21.4 0.103
ALT (U/L) 27.3 ± 17.9 27.8 ± 26.0 0.878
AST (U/L) 21.1 ± 10.6 21.8 ± 12.2 0.677
LAD (cm) 4.1 ± 2.6 3.9 ± 0.5 0.443
EF (%) 61.2 ± 5.0 60.4 ± 8.0 0.397
Follow-Up duration (days) 344.0 ± 60.5 346.8 ± 59.9 0.737
Procedure duration (minutes) 155.6 ± 46.1 157.8 ± 67.4 0.776
Fluoroscopy time (minutes) 10.5 ± 5.9 9.9 ± 6.0 0.460
Radiation dose (mGy) 245.7 ± 173.9 229.9 ± 178.5 0.511
BMI indicates body mass index; Scr, serum creatinine; GFR, glomerular filtration rate; ALT,
alanine transaminase; AST, aspartate transaminase; LAD, left atrial diameter; and EF, ejection
fraction. Continuous variables were compared using the two-tailed Student’s t-test, and cate-
gorical variables were compared using the χ2 test or Fischer’s exact test.
recommended after the “blanking period” when the CHA2 higher in the BT Group (25/49, 51.0%) than in the TF
DS2-VASc score > 1. Patients who regularly took oral an- Group (16/63, 25.4%, P = 0.0052). There were 11/109
ticoagulation covering the “blanking period” were defined (10.1%) and 2/109 (1.8%) patients receiving a second ab-
as adherence to anticoagulation, and it was also recorded. lation who suffered from AF recurrence, atrial tachycardia,
Statistical analysis: Continuous variables are expressed as and atrial flutter in BT Group and TF Group, respectively
mean ± standard deviation, while categorical variables are (P = 0.0101).
expressed as number and percentage. Continuous variables The Kaplan-Meier curves for recurrence stratified by
were compared using the two-tailed Student’s t-test, follow-up strategies at the time of enrollment are shown
whereas categorical variables were compared using the χ2 in Figure 2. Log-rank tests showed a significantly earlier
test or Fisher’s exact test. In order to examine the influ- documentation of AF recurrence in the BT Group than in
ence of follow-up strategies on AF recurrence detection, the TF Group (P < 0.05). In a multivariate Cox model,
Kaplan-Meier survival curve and log-rank analysis were the likelihood of recurrence detection was more significant
performed. For Cox regression, univariable regression was in patients in the BT Group with larger left atrium (Table
performed first. Variables with a P-value of < 0.2 were II).
then included in the multivariable model. A level of sig- Feasibility of BigThumb in follow-up after ablation: In
nificance was set at < 0.05 for all reported P-values, and total, 26133 bECGs were recorded, among which 3299
confidence intervals were calculated at the 95% level. All (12.6%) were confirmed as AF by cardiologists’ manual
statistics were then calculated using SPSS (version 24, review, 3860 (14.8%) by the automated AF detection al-
IBM, Armonk, NY, USA). gorithm, and 3457 (13.2%) by the AI algorithm. The re-
corded bECGs were 239.8 ± 87.6/patients, while AF
bECGs were 30.3 ± 11.7/patients. The sensitivity of the
Results
AI algorithm was significantly higher than the automated
Baseline characteristics: In total, 109 patients were in- AF detection algorithm [(94.4%, 95% CI, 93.5-95.1%)
cluded in each group. No statistical difference was de- versus (90.7%, 95% CI, 89.7%-91.7%)]. The specificity
tected in the demographic data between patients in the was also higher in the AI algorithm [(98.5%, 95% CI,
two groups (Table I). 98.3%-98.6%) versus (96.2%, 95% CI, 95.9%-96.4%)].
Primary outcome: After follow-up of 344.0 ± 60.5 days There were 2514/3299 (76.2%) AF bECGs marked with
in the BT Group, 70/109 patients (64.2%) were free from symptoms by participants.
AF recurrence (70.8% in paroxysmal AF and 54.5% in Compliance of BigThumb monitoring: The monitoring
persistent AF). While in the TF Group, the AF recurrence- frequency was 1.71 ± 0.02/day in all the participants.
free rate was 86/109 patients (78.9%, P = 0.0163) after They were used more frequently in the first 3 months of
follow-up of 346.8 ± 59.9 days (88.2% in paroxysmal AF follow-up than after (2.73 ± 0.03/day versus 1.29 ± 0.02/
and 57.6% in persistent AF). The BigThumb detected day, P < 0.05). The monitoring was most frequently re-
more AF recurrence in paroxysmal AF after ablation (P = corded in the daytime than in the nighttime. Hence, the
0.0099), but not in persistent AF (P = 0.7910). episodes of AF were more frequently detected during day-
Adherence to anticoagulation in patients with CHA2 time (Figure 3). In the TF Group, meanwhile, Holter rate
DS2-VASc score > 1 was determined to be significantly was 3.2 ± 0.8 per patient.Int Heart J
July 2021 PORTABLE DEVICE IMPROVES ATRIAL FIBRILLATION DETECTION 789
Figure 2. The Kaplan–Meier survival curve for the primary endpoint. AF indicates atrial
fibrillation; TF, traditional follow-up; and BT, BigThumb.
Table II. Cox Regression Analysis with AF Recurrence
Univariable analysis Multivariable analysis
Variable
Β-Coefficient (95% CI) P Β-Coefficient (95% CI) P
Follow-up strategy 0.33 (0.15–0.73) 0.006 0.41 (0.18–0.91) 0.029
LAD 2.79 (1.60–4.89) 0.001 2.34 (1.20–4.57) 0.013
LAD indicates left atrial diameter. For Cox regression, univariable regression was performed first.
Variables with a P-value of < 0.2 were included in the multivariable model (follow-up strategy and
LAD). A level of significance was set at < 0.05 for all reported P-values, and confidence intervals
were calculated at the 95% level.
AliveCor is a monitor attached to a WiFi-enabled
Discussion
iPod to obtain ECGs in ambulatory patients.15) It will defi-
To the best of our knowledge, this is the first study nitely make great progress in AF detection with its con-
to demonstrate the feasibility and veracity of the textualized medical-grade ECG technology brought to the
BigThumb monitoring device. The primary finding of this patients, as compared with the current practice.16) Halcox,
study is that follow-up after AF ablation using BigThumb et al. conducted a randomized controlled trial of AF
leads to a more frequent detection of AF recurrence and screening using AliveCor to obtain ECGs.17) Among the
more adherence to oral anticoagulation. AI algorithm has 60440 ECG tracings, only 1% (600 ECGs) were catego-
been found to improve the accuracy of ECG diagnosis. rized as AF by the automated AliveCor algorithm. Of
It was reassuring to see abundant devices and appli- these, only 5% (30 tracings) were confirmed to be AF.
cations available in post-ablation follow-up. ECG and The low positive predictive value of the AliveCor for the
Holter monitoring are the most frequently applied tech- detection of AF was an unexpected observation, making
niques to detect AF recurrence in previous studies. How- the current generation of AliveCor devices as a screening
ever, low-frequency monitoring is not sufficient for tool difficult to justify.18) Similarly, the BigThumb, which
follow-up.7,8) Pulse wave monitor devices are also used in is used for smartphone-based ECG collection, paralleled
AF detection, but with unsatisfied accuracy.9,10) the AliveCor in structure design, but the sensitivity and
Accumulating evidence and meta-analyses have dem- specificity of the AI algorithm were 94.4% and 98.5%, re-
onstrated that insertable cardiac monitor (ICM) and car- spectively. It has the potential to reduce the manual re-
diac implantable electronic device detect a high rate of AF view. The use of mobile ECG self-recording devices al-
typically missed during routine clinical care in patients af- lows for earlier detection of AF recurrence and may em-
ter ablation.11-13) Early detection of AF has been identified power patients to engage in shared health decision mak-
to be crucial in order to define or change proper medical ing.
treatment.14) However, implantation of the devices is inva- There is emerging evidence that AF recurrence after
sive and expensive, which prevent them from the popular- ablation may be asymptomatic. Studies have shown that
ity in follow-up. asymptomatic AF may be increasingly common after abla-Int Heart J
790 HUANG, ET AL July 2021
cisions after ablation guided by the BigThumb. High-
quality research is needed to understand the efficacy of
coagulation strategies guided by the BigThumb.
Disclosure
Conflicts of interest: The authors declare that no con-
flicts of interest.
Ethics approval and consent to participate: The study
was approved by the ethics committee of Changhai Hospi-
tal (No. 20190903). All patients provided written informed
consent.
Availability of data and materials: The deidentified par-
ticipant data will not be shared.
Figure 3. The monitor frequency collected in 24-hour duration.
References
tion, thus potentially posing a greater thromboembolic risk 1. Hindricks G, Potpara T, Dagres N, et al. ESC guidelines for the
than symptomatic AF.19-21) In this present study, we found diagnosis and management of atrial fibrillation developed in
collaboration with the European Association for Cardio-
that 785/3299 (23.8%) of the AF detected after ablation
Thoracic Surgery (EACTS): The task force for the diagnosis
were asymptomatic, together with unawareness of AF- and management of atrial fibrillation of the European Society of
related outcomes and consequent comorbidities, which Cardiology (ESC) developed with the special contribution of the
may explain the poor compliance with anticoagulation af- European Heart Rhythm Association (EHRA) of the ESC. Eur
ter catheter ablation even if patients receive recommenda- Heart J 2021; 42: 373-498.
tions of anticoagulation from doctors. The BigThumb may 2. Chen LY, Chung MK, Allen LA, et al. Atrial fibrillation burden:
help to record the episode of asymptomatic AF according Moving beyond atrial fibrillation as a binary entity: A scientific
statement from the American Heart Association. Circulation
to random monitoring, which may lead to subsequent ad- 2018; 137: e623-44.
herence to anticoagulation. 3. Andrade JG, Champagne J, Dubuc M, et al. Cryoballoon or ra-
The BigThumb is an effective and affordable public diofrequency ablation for atrial fibrillation assessed by continu-
ECG monitoring device, which increase the success of in- ous monitoring: A randomized clinical trial. Circulation 2019;
corporating the consumer-generated biometrics into clini- 140: 1779-88.
cal practice. However, it has limitations. ECG tracings in 4. Sanna T, Diener HC, Passman RS, et al. Cryptogenic stroke and
underlying atrial fibrillation. N Engl J Med 2014; 370: 2478-86.
this clinical trial provide an insight into the real-world
5. Turakhia MP, Desai M, Hedlin H, et al. Rationale and design of
limitations of the technology. Because of random and hap- a large-scale, app-based study to identify cardiac arrhythmias
hazard collection of bECGs, the accuracy of AF detection using a smartwatch: The Apple Heart study. Am Heart J 2019;
with the BigThumb highly depends on the compliance of 207: 66-75.
participants. As we have found in this study, the 6. Guo Y, Wang H, Zhang H, et al. Mobile photoplethysmographic
BigThumb has been used frequently in the first 3-month technology to detect atrial fibrillation. J Am Coll Cardiol 2019;
follow-up than after three months. Patients should be re- 74: 2365-75.
7. Yushan B, Tan BYQ, Ngiam NJ, et al. Association between bi-
minded to persistently utilize the device. We also found in lateral infarcts pattern and detection of occult atrial fibrillation
this study that the frequency of monitoring in the night in embolic stroke of undetermined source (ESUS) patients with
was significantly lower than the daytime as expected. insertable cardiac monitor (ICM). J Stroke Cerebrovasc Dis
Some asymptomatic episodes in the night may be missed. 2019; 28: 2448-52.
Moreover, there is potential for heightened concern with 8. Fukuma N, Hasumi E, Fujiu K, et al. Feasibility of a t-shirt-type
abnormal or equivocal readings in spite of the best inten- wearable electrocardiography monitor for detection of covert
atrial fibrillation in young healthy adults. Sci Rep 2019; 9:
tions of a health-conscious and motivated individual.
11768.
In conclusion, we found that follow-up after AF abla- 9. Seshadri DR, Bittel B, Browsky D, et al. Accuracy of the apple
tion using BigThumb with AI algorithm is more effective Watch 4 to measure heart rate in patients with atrial fibrillation.
than traditional strategies. It is conceivable that in the near IEEE J Transl Eng Health Med 2020; 8: 2700204.
future, most patients after ablation will be continuously 10. Bonomi AG, Schipper F, Eerikäinen LM, et al. Atrial fibrillation
and indefinitely monitored. detection using a novel cardiac ambulatory monitor based on
Limitations: This study has limitations. First, this study is photo-plethysmography at the wrist. J Am Heart Assoc 2018; 7:
e009351.
a single-center clinical trial with limited follow-up period 11. Philippsen TJ, Christensen LS, Hansen MG, Dahl JS, Brandes
and comparatively small simple size that may introduce A. Detection of subclinical atrial fibrillation in high-risk patients
patient selection bias. Second, the clinical setting was spe- using an insertable cardiac monitor. JACC Clin Electrophysiol
cific to China, so our results should be understood in that 2017; 3: 1557-64.
context. Finally, because there were few episodes of 12. Ciconte G, Saviano M, Giannelli L, et al. Atrial fibrillation de-
thromboembolism or bleeding, our study does not have tection using a novel three-vector cardiac implantable monitor:
The atrial fibrillation detect study. Europace 2017; 19: 1101-8.
sufficient power to detect a difference with the low rate of
13. Nölker G, Mayer J, Boldt LH, et al. Performance of an implant-
complications observed according to different clinical de-Int Heart J
July 2021 PORTABLE DEVICE IMPROVES ATRIAL FIBRILLATION DETECTION 791
able cardiac monitor to detect atrial fibrillation: Results of the AliveCor Heart monitor to screen for atrial fibrillation: The
DETECT AF study. J Cardiovasc Electrophysiol 2016; 27: REHEARSE-AF study”. Circulation 2018; 137: 2191-2.
1403-10. 19. Scacciatella P, Jorfida M, Biava LM, et al. Insertable cardiac
14. Wasser K, Weber-Krüger M, Jürries F, et al. The cardiac diag- monitor detection of silent atrial fibrillation in candidates for
nostic work-up in stroke patients-A subanalysis of the Find- percutaneous patent foramen ovale closure. J Cardiovasc Med
AFRANDOMISED trial. PLOS ONE 2019; 14: e0216530. (Hagerstown) 2019; 20: 290-6.
15. Chan PH, Wong CK, Pun L, et al. Head-to-head comparison of 20. Iwata T, Todo K, Yamagami H, et al. High detection rate of
the AliveCor heart monitor and microlife WatchBP office AFIB atrial fibrillation with insertable cardiac monitor implantation in
for atrial fibrillation screening in a primary care setting. Circu- patients with cryptogenic stroke diagnosed by magnetic reso-
lation 2017; 135: 110-2. nance imaging. J Stroke Cerebrovasc Dis 2019; 28: 2569-73.
16. Verma A, Wachter R, Kowey PR, et al. Changes in management 21. Goldenthal IL, Sciacca RR, Riga T, et al. Recurrent atrial fibril-
following detection of previously unknown atrial fibrillation by lation/flutter detection after ablation or cardioversion using the
an insertable cardiac monitor (from the REVEAL AF study). AliveCor KardiaMobile device: iHEART results. J Cardiovasc
Am J Cardiol 2019; 124: 864-70. Electrophysiol 2019; 30: 2220-8.
17. Halcox JPJ, Wareham K, Cardew A, et al. Assessment of remote
heart rhythm sampling using the AliveCor heart monitor to
screen for atrial fibrillation: The REHEARSE-AF study. Circu- Supplemental Files
lation 2017; 136: 1784-94.
Supplemental Text
18. Koshy AN, Sajeev JK, Teh AW. Letter by Koshy et al regarding
article, “Assessment of remote heart rhythm sampling using the Please see supplemental files; https://doi.org/10.1536/ihj.21-067You can also read
