Original Research Articles: Original Clinical Research Report

Wireless Single-Lead ECG Monitoring to Detect New-Onset Postoperative Atrial Fibrillation in Patients After Major Noncardiac Surgery: A Prospective Observational Study

Jokinen, Johan D. V. MD^*,†; Carlsson, Christian J. MD, PhD^*,†; Rasmussen, Søren M. MSc^‡; Nielsen, Olav W. MD, DMSc, PhD^§; Winkel, Bo G. MD, PhD^‖; Jorgensen, Lars N. MD, DMSc^¶; Achiam, Michael P. MD, DMSc, PhD^#; Mølgaard, Jesper MD^**; Sørensen, Helge B. D. MSc, PhD^*; Aasvang, Eske K. MD, DMSc^**,††; Meyhoff, Christian S. MD, PhD^*,†,††; for the WARD Project Group

Author Information

From the ^*Department of Anaesthesia and Intensive Care, Copenhagen University Hospital - Bispebjerg and Frederiksberg, Copenhagen, Denmark

^†Copenhagen Center for Translational Research, Copenhagen University Hospital - Bispebjerg and Frederiksberg, Copenhagen, Denmark

^‡Digital Health Section, Department of Health Technology, Technical University of Denmark, Kgs-Lyngby, Denmark

^§Department of Cardiology, Copenhagen University Hospital - Bispebjerg and Frederiksberg, Copenhagen, Denmark

^‖Department of Cardiology, The Heart Centre, Copenhagen University Hospital-Rigshospitalet, Copenhagen, Denmark

^¶Digestive Disease Center K, Copenhagen University Hospital - Bispebjerg and Frederiksberg, Copenhagen, Denmark

^#Department of Surgical Gastroenterology, Centre for Cancer and Organ Diseases, Copenhagen University Hospital - Rigshospitalet, Copenhagen, Denmark

^**Department of Anaesthesiology, Centre for Cancer and Organ Diseases, Copenhagen University Hospital - Rigshospitalet, Copenhagen, Denmark

^††Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark.

Published ahead of print February 25, 2022.

Accepted for publication December 14, 2021.

Funding: The Wireless Assessment of Respiratory and circulatory Distress (WARD) project received core support from the Innovation Fund Denmark (8056-00055B), the Novo Nordic Foundation, the Danish Cancer Society (R150-A9865-16-S48), Steno Diabetes Center Denmark, Copenhagen Center for Health Technology (CACHET), Radiometer Medical Aps, Isansys Ltd, A. P. Møller Foundation, Bispebjerg Hospital, Rigshospitalet, and the Technical University of Denmark. No sponsor had any role in the study design, in the collection, analysis, and interpretation of data, nor in writing the report or the decision to submit the article. No sponsor had access to the study data.

Conflicts of Interest: See Disclosures at the end of the article.

Reprints will not be available from the authors.

Clinical Trial numbers: NCT03491137 (https://clinicaltrials.gov/ct2/show/NCT03491137).

H. B. D. Sørensen, E. K. Aasvang, and C. S. Meyhoff shared last authorship.

Address correspondence to Johan D. V. Jokinen, MD, Department of Anesthesia and Intensive Care, Bispebjerg and Frederiksberg Hospital, University of Copenhagen, Bispebjerg Bakke 23, 2400 Copenhagen, NV, Denmark. Address e-mail to [email protected].

Anesthesia & Analgesia 135(1):p 100-109, July 2022. | DOI: 10.1213/ANE.0000000000005960

Free

Abstract

BACKGROUND:

New-onset postoperative atrial fibrillation (POAF) is associated with several cardiovascular complications and higher mortality. Several pathophysiological processes such as hypoxia can trigger POAF, but these are sparsely elucidated, and POAF is often asymptomatic. In patients undergoing major gastrointestinal cancer surgery, we aimed to describe the frequency of POAF as automatically estimated and detected via wireless repeated sampling monitoring and secondarily to describe the association between preceding vital sign deviations and POAF.

METHOD:

Patients ≥60 years of age undergoing major gastrointestinal cancer surgery were continuously monitored for up to 4 days postoperatively. Electrocardiograms were obtained every minute throughout the monitoring period. Clinical staff were blinded to all measurements. As for the primary outcome, POAF was defined as 30 consecutive minutes or more detected by a purpose-built computerized algorithm and validated by cardiologists. The primary exposure variable was any episode of peripheral oxygen saturation (Spo₂) <85% for >5 consecutive minutes before POAF.

RESULTS:

A total of 30,145 hours of monitoring was performed in 398 patients, with a median of 92 hours per patient (interquartile range [IQR], 54–96). POAF was detected in 26 patients (6.5%; 95% confidence interval [CI], 4.5–9.4) compared with 14 (3.5%; 95% CI, 1.94–5.83) discovered by clinical staff in the monitoring period. POAF was followed by 9.4 days hospitalization (IQR, 6.5–16) versus 6.5 days (IQR, 2.5–11) in patients without POAF. Preceding episodes of Spo₂ <85% for >5 minutes (OR, 1.02; 95% CI, 0.24-4.00; P = .98) or other vital sign deviations were not significantly associated with POAF.

CONCLUSIONS:

New-onset POAF occurred in 6.5% (95% CI, 4.5–9.4) of patients after major gastrointestinal cancer surgery, and 1 in 3 cases was not detected by the clinical staff (35%; 95% CI, 17–56). POAF was not preceded by vital sign deviations.

KEY POINTS

Questions: Can new-onset postoperative atrial fibrillation (POAF) after major abdominal surgery be detected in the surgical wards by continuous wearable monitoring, and is POAF preceded by deviating vital signs?
Findings: A high detection rate of POAF (6.5%, 95% confidence interval [CI], 4.5–9.4) was found using wireless monitoring with repeated sampling, and no significant associations were found between POAF and preceding deviating vital signs.
Meaning: Wireless repeated sampling monitoring can detect POAF outside routine monitoring in surgical wards, and these events do not seem to be preceded by vital sign derangement.

Atrial fibrillation (AF) is the most common cardiac arrhythmia after surgery.^¹^,^² New-onset postoperative AF (POAF) can be associated with several cardiovascular complications,^³^,^⁴ higher mortality, prolonged hospitalization, and increased health care expenditures.^{^5–7} The frequency of POAF after noncardiac surgery has been reported to be between 0.4% and 4.4%,^{^5–10} depending on population, type of surgery, and type of detection tool. However, POAF is underestimated as detection often relies on intermittent manual assessments by clinical staff, when patients are outside high-dependency wards or telemetry units.^¹¹ Activation of the sympathetic nervous system and the acute stress response induced by surgery are important triggers of POAF,^¹² yet the pathophysiology is multifactorial and often a result of several concurrent processes. Perioperative risk factors for POAF include anemia, hypovolemia and hypervolemia, hypotension, hypoxia, electrolyte imbalances, inflammation, and metabolic derangements.^¹² Similarly, acute deviations of vital signs are detected at intervals of up to 12 hours^¹¹ and may therefore precede and trigger POAF. Recent medicotechnical advances have allowed for wireless close monitoring of heart rhythm and vital signs with repeated sampling technique during postoperative recovery in the general surgical wards. This may identify POAF with higher sensitivity and detect preceding deviating vital signs,^¹³^,^¹⁴ thus allowing evidence-based treatment for all patients with POAF. We hypothesized that algorithmic estimation of AF via wireless single-lead electrocardiogram (ECG) monitoring can detect POAF frequently after major gastrointestinal cancer surgery compared with previous studies using routine monitoring outside telemetry units. This study aimed to describe the frequency of new-onset POAF defined as patients having 1 or more episodes lasting at least 30 minutes as measured continuously with a wearable single-lead ECG device by repeated sampling. Secondarily, we described the associations between deviating vital signs and subsequent development of POAF.

METHODS

This study is nested in the prospective, observational cohort study Wireless Assessment of Respiratory and circulatory Distress (WARD; clinicaltrials.gov record NCT03491137). The study was approved by the Danish Data Protection Agency (RH-2016-390) and the regional ethics committee (H-17033535) on February 8, 2018. This study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Patients

Patients undergoing elective major gastrointestinal cancer surgery from February 2018 to June 2020 at Copenhagen University hospitals Bispebjerg Hospital and Rigshospitalet, Denmark were included. Eligible patients were ≥60 years of age with an expected duration of surgery of ≥2 hours and at least 1 expected overnight stay. Patients were excluded if they were not expected to be able to comply with study procedures, unable to provide informed consent, had a pacemaker or implantable cardioverter defibrillator (ICD) unit, a Mini-Mental State Examination score <24,^¹⁵ or if allergic to plaster, silicone, or plastic. Patients with chronic AF, recent AF, or presenting with AF on preoperative ECG were also excluded. Written informed consent was obtained preoperatively, and monitoring was started postoperatively in the postanesthesia care unit (PACU) and onward for up to 4 days or until hospital discharge.

Monitoring and Data Collection

Patients were monitored continuously using 3 wireless devices: (1) the Isansys LifeTouch single-lead ECG sensor (Isansys Lifecare) is a patch with 2 electrodes 15 cm apart placed over the anterior left aspect of the patient’s thorax at an angle of 45° with its base in the fourth intercostal space approximately 2 cm lateral of sternum. It continuously records ECG signals from which repeated sampling of measurements of heart rhythm, heart rate (HR), and respiratory rate (RR) are derived. (2) The Nonin WristOx 3150 wrist-worn pulse oximeter (Nonin Medical Inc) measures pulse rate and peripheral oxygen saturation (Spo₂) with a sampling frequency of 1 per second. (3) The Meditech BlueBP-05 blood pressure monitor (Meditech Ltd) measures a noninvasive systolic blood pressure (SBP) and diastolic blood pressure (DBP) at 30-minute intervals during the day and evening and 1-hour intervals during nighttime. Data from the Isansys LifeTouch sensor and the Nonin pulse oximeter were automatically transmitted in real time via Bluetooth to a password-protected tablet device kept next to the patient’s hospital bed. Ten seconds of ECG recording every minute and 1-minute averages of HR, RR, and Spo₂ were uploaded and stored on a secure server in real time. Data from the Meditech blood pressure monitor were stored on the unit for daily manual transfer of data to the server. In cases of missing data, measurements of HR, RR, Spo₂, and blood pressure measurements were extrapolated for no more than 60 minutes. Clinical staff and patients were blinded to all collected vital sign data throughout hospitalization.

Vital Signs Data Processing and AF Detection Algorithm

Artifacts were detected and removed before data analysis: an absolute change in Spo₂ >4% within 1 second was considered an artifact.^¹⁶ The presence of noise in the ECG signal was detected using an algorithm applied to 10-second segments of ECG recording every minute of monitoring. Quality of signals was evaluated, and artifacts were considered in case of low amplitude or extremes of HR. Detection of AF relied on a purpose-built, computerized signal processing algorithm based on ECG recordings in patients hospitalized with verified AF. R-wave peak intervals were analyzed automatically for regularity in 1-minute ECG signals to determine the probability of AF and were identified as potential AF if present for 30 consecutive minutes. After all data from the study were collected, POAF was registered only if adjudication by 2 cardiology specialists (O.W.N. and B.G.W.) blinded to information from medical records confirmed AF during the final 5 minutes of ECG recording leading up to potential AF identification. This annotation of AF versus non-AF segment is used to train the POAF algorithm in the WARD project, which will be a real-time 24-hour, 7-day analysis and alerting system and implemented in future studies. The algorithm was not built to identify occurrence of atrial flutter. Therefore, the primary outcome was the frequency of POAF lasting at least 30 consecutive minutes and occurring during the first 4 days of monitoring after surgery.

Exposure Variables

Thresholds for severity and duration of deviating vital signs (Spo₂, RR, HR, and SBP) were defined a priori as microevents (Table 1). The number of episodes exceeding these thresholds in the 24 hours before the onset of POAF or in the first 24 hours of monitoring time in patients who did not develop POAF was used as exposure variables. The primary exposure variable was an episode with an Spo₂ <85% for at least 5 consecutive minutes.

Statistical Analyses

This study was an additional analysis of the WARD surgical study (NCT03491137), in which the sample size of the 500 patients was based on the requirement of 100 patients with postoperative complications to allow machine learning-based algorithms and approximately 20% occurrence of serious adverse events in major abdominal surgery. Baseline variables in patients with and without POAF as well as occurrence and characteristics of POAF summarized as pooled estimates of the recorded events were presented in frequency tables, with time-to-onset timed at the first occurrence and duration as averaged of the cumulated period of POAF. The association of baseline variables and episodes of deviating vital signs with subsequent development of POAF was analyzed using the Fisher exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. Independent risk factors for POAF were identified using stepwise backward selection by Akaike information criterion before being included in a multivariable regression model ultimately adjusted for age, body mass index (BMI), smoking, alcohol, American Society of Anesthesiologists (ASA) score, paroxysmal AF, myocardial infarction, aorta stenosis, hypercholesterolemia, diabetes mellitus, open surgery, gastric cancer, pancreatic cancer, rectum cancer, preoperative creatinine, SBP, and ECG monitoring hours. Variables with a P value of <.05 were entered into a logistic regression model analyzing the association of episodes of deviating vital signs and POAF. Unadjusted odds ratios (ORs) as well as adjusted ORs (primary results) were reported. No adjustments for multiple comparisons were performed as results were exploratory. Hospital length of stay and mortality were analyzed with unadjusted Wilcoxon rank sums test. A P value <.05 was considered significant. All statistical analyses were performed using the statistical software R (v.3.6.2).^¹⁷

RESULTS

A total of 468 patients entered this study, of whom 70 patients were excluded, leaving a total sample of 398 patients (Figure 1). The median age was 70 (interquartile range [IQR], 66–74) years; 59% were men, the most common comorbidities were anemia (61%) and hypertension (47%), and 61% underwent open surgery (Table 2). We collected a total of 30,145 hours of wireless patient monitoring (26,480 hours of ECG recordings, 17,363 hours of Spo₂, and 17,015 hours of blood pressure measurement), with a median monitoring time of 92 hours per patient (IQR, 54–96). Median duration of ECG monitoring in patients with POAF was 82 hours (IQR, 66–94) versus 71 hours (IQR, 45–93) in patients without POAF (P = .034), and duration of peripheral oximetry and blood pressure monitoring was not significantly different (Table 2).

Table 1. - Vital Sign Deviations in the 24 Hours Before Onset of POAF or During the First 24 Hours of Monitoring in Patients Without POAF After Major Gastrointestinal Cancer Surgery

Microevent	POAF	No POAF	Unadjusted OR (95% CI)	P value	Adjusted OR (95% CI)	P value
Microevent	No. (%)	No. (%)	Unadjusted OR (95% CI)	P value	Adjusted OR (95% CI)	P value
Hypoxemia
Spo ₂ <92% ≥60 min	9 (35)	159 (43)	0.70 (0.27–1.72)	.54	0.69 (0.22–1.98)	.50
Spo ₂ <88% ≥10 min	9 (35)	142 (38)	0.86 (0.33–2.10)	.84	0.78 (0.18–2.99)	.72
Spo ₂ <85% ≥5 min	8 (31)	117 (32)	0.97 (0.35–2.42)	1.00	1.02 (0.24–4.00)	.98
Spo ₂ <80% ≥1 min	4 (15)	66 (18)	0.84 (0.20–2.60)	1.00	1.47 (0.35–5.23)	.57
Bradypnea
RR <6·min⁻¹ ≥1 min	2 (7.7)	36 (9.7)	0.77 (0.09–3.36)	1.00	0.49 (0.05–2.53)	.47
Tachypnea
RR ≥25·min⁻¹ ≥5 min	9 (35)	61 (16)	2.59 (0.97–6.48)	.03	1.71 (0.53–4.84)	.33
Hypoventilation
RR <11 and Spo ₂ <88% ≥5 min	1 (3.8)	17 (4.6)	0.84 (0.02–5.79)	1.00	1.48 (0.07–11)	.74
Bradycardia
HR <41 bpm ≥5 min	1 (3.8)	3 (0.8)	4.88 (0.09–63)	.24	9.4 (0.35 to >99)	.11
Tachycardia
HR >110 bpm ≥60 min	2 (7.7)	11 (3.0)	2.72 (0.28–13)	.21	0 (NA to >99)	.99
HR >130 bpm ≥60 min	3 (12)	1 (0.3)	47 (3.60 to >99)	-	>99 (<0.01 to NA)	.99
Hypotension
SBP <70 mm Hg ≥1 measurement	0 (0.0)	2 (0.5)	0 (<0.01–77)	1.00	0 (NA to >99)	.99
SBP <91 mm Hg ≥60 min	3 (12)	74 (20)	0.53 (0.10–1.81)	.44	0.30 (0.05–1.10)	.11
Hypertension
SBP >180 mm Hg ≥60 min	1 (3.8)	28 (7.6)	0.49 (0.01–3.25)	.71	0 (NA to >99)	.99
SBP >220 mm Hg ≥1 measurement	1 (3.8)	2 (0.5)	7.32 (0.12–145)	.18	>99 (NA to >99)	.99

Exposure variables are patients with 1 or more vital sign deviations. Estimates for unadjusted ORs are calculated with the Fisher exact test, and adjusted OR estimates are calculated with a logistic regression model with mutual adjustment and adjustment for paroxysmal atrial fibrillation and duration of ECG monitoring.

Abbreviations: CI, confidence interval; HR, heart rate; NA, not applicable; OR, odds ratio; POAF, postoperative atrial fibrillation; RR, respiratory rate; SBP, systolic blood pressure; Spo₂, peripheral oxygen saturation.

Table 2. - Characteristics of Patients Undergoing Major Gastrointestinal Cancer Surgery

Characteristic	POAF	No POAF	Total
Characteristic	n = 26 (6.5%)	n = 372 (93.5%)	N = 398 (100%)
Male, No. (%)	15 (58)	222 (60)	237 (60)
Age, median (IQR)	74 (68–76)	70 (66–74)	70 (66–74)
Body mass index, median (IQR)	24 (21–25)	25 (23–28)	25 (23–28)
Smoking status, No. (%)
Never	11 (42)	122 (33)	133 (33)
Former	10 (39)	197 (53)	207 (52)
Current	5 (19)	53 (14)	58 (15)
Alcohol, No. (%)
None	9 (35)	79 (21)	88 (22)
Within recommended weekly intake^{^a}	14 (54)	225 (61)	239 (60)
Above recommended weekly intake^{^a}	3 (12)	68 (18)	71 (18)
ASA score, No. (%)
I	0	19 (5.1)	19 (4.8)
II	14 (54)	208 (56)	222 (56)
III	11 (42)	143 (38)	154 (39)
IV	1 (3.8)	2 (0.5)	3 (0.8)
Comorbidities, No. (%)
Stroke or transient ischemic attack	1 (3.8)	24 (6.5)	25 (6.3)
Paroxysmal atrial fibrillation	6 (23)	12 (3.2)	18 (4.5)
Myocardial infarction	3 (12)	20 (5.4)	23 (5.8)
Aorta stenosis	2 (7.7)	5 (1.3)	7 (1.8)
Congestive heart failure	1 (3.8)	5 (1.3)	6 (1.5)
Pulmonary embolism	0	6 (1.6)	6 (1.5)
Chronic obstructive pulmonary disease	3 (112)	25 (6.7)	28 (7.0)
Asthma	2 (7.7)	13 (3.5)	15 (3.8)
Hypertension	15 (58)	173 (47)	188 (47)
Hypercholesterolemia	4 (15)	79 (21)	83 (21)
Diabetes mellitus	5 (19)	58 (16)	63 (16)
Anemia^{^b}	13 (57)	149 (51)	181 (49)
Open surgery, No. (%)	17 (65)	224 (60)	241 (61)
Type of cancer, No. (%)
Esophageal	6 (23)	57 (15)	63 (16)
Gastric	2 (7.7)	44 (12)	46 (12)
Duodenal or small bowel	0	30 (8.1)	31 (7.8)
Pancreatic	9 (35)	141 (38)	152 (38)
Colonic	6 (23)	95 (26)	101 (25)
Rectal	1 (3.8)	34 (9.1)	35 (8.8)
Preoperative laboratory values and vital signs
Hemoglobin, median (IQR) mmol/L	7.9 (7.1–8.2)	7.9 (7.1–8.2)	7.9 (7.1–8.6)
Creatinine, median (IQR) mmol/L	82 (68–103)	77 (67–90)	77 (67–90)
Hs-TnT, median (IQR) ng/L^{^c}	28 (22–35)	25 (23–39)	25 (23–37)
SBP, median (IQR) mm Hg	134 (124–146)	138 (125–149)	136 (125–149)
DBP, median (IQR) mm Hg	77 (70–85)	76 (68–83)	76 (69–84)
Spo ₂, median (IQR) %	98 (97–99)	98 (97–99)	98 (97–99)
Postoperative patient monitoring, median (IQR) hours
Single-lead ECG	82 (66–94)	71 (45–93)	72 (46–93)
Peripheral oximetry	50 (39–70)	42 (20–66)	43 (20–67)
Blood pressure	44 (22–73)	43 (19–65)	43 (19–65)

^aAs recommended by the Danish Health Authority: 24 g/d for men or 12 g/d for women.

^bDefined as hemoglobin <8.3 mmol/L for men or <7.3 mmol/L for women.

^cValues from patients with Hs-TnT ≥20 ng/L only.

Abbreviations: ASA, American Society of Anesthesiologists; DBP, diastolic blood pressure; ECG, electrocardiogram; Hs-TnT, high-sensitive troponin-T; IQR, interquartile range; POAF, postoperative atrial fibrillation; SBP, systolic blood pressure; Spo₂, peripheral oxygen saturation.

Figure 1.:
Study flowchart. COVID-19 indicates coronavirus disease 2019; ECG, electrocardiogram; POAF, postoperative atrial fibrillation.

The primary outcome of POAF lasting ≥30 consecutive minutes was confirmed in 26 of the 398 patients (6.5%; 95% confidence interval [CI], 4.5–9.4). There was a total of 108 periods with AF registered across these patients (median 2.5 periods per patient with POAF; IQR, 1.00–4.75), and the median duration of POAF was 82 minutes(IQR, 55–164) (Table 3). POAF most frequently developed postoperatively on day 2 (35%; 95% CI, 17–56) and day 3 (35%; 95% CI, 17–56) (Figure 2). A total of 17 patients (65%; 95% CI, 44–83) recovered to sinus rhythm without cardioversion or antiarrhythmic agents before the monitoring ended (Table 3). Median HR during POAF episodes was 100 bpm (range, 53–195). Of the 26 patients with POAF, 14 patients were detected clinically by clinical staff during the monitoring period (3.5% of the study cohort; 95% CI, 1.9–5.8). Two patients had self-limiting POAF of short duration that was not detected by the monitoring equipment (false negatives).

Table 3. - Pooled Estimates of the Events of New-Onset POAF During Admission to General Surgical Wards After Major Gastrointestinal Cancer Surgery

Characteristic	POAF	POAF detected by clinical staff
	POAF	During monitoring	Between monitoring and discharge	Never detected
	n = 26 (6.5%)	n = 14 (3.5%)	n = 3 (0.8%)	n = 9 (2.3%)
Time to onset, median (IQR) hours	49 (38–61)	49 (38–62)	80 (45–114)	49 (37–59)
Heart rate during POAF, median (range) beats per minute	100 (53–195)	106 (49–196)	96 (63–163)	77 (42–170)
Duration of POAF, median (IQR) minutes	82 (55–164)	126 (61–218)	210 (139–397)	62 (41–80)
Frequency of reconversion, No. (%)	19 (73)	10 (71)	2 (67)	7 (78)
Without intervention, No. (%)	17 (65)	8 (14)	1 (33)	7 (78)
Antiarrhythmic treatment initiated, No (%)	2 (8)	2 (14)	0	0
Frequency of no reconversion, No. (%)	7 (27)	4 (29)	1 (33)	2 (22)
Antiarrhythmic treatment initiated, No (%)	0	0	0	0
Hospital length of stay, median (IQR) days	9.4 (6.5–16)	9.4 (7.4–16)	15 (6–31)	6.5 (2.9–19)

Detection of reconversion included inspection of the final 5 minutes of ECG recordings.

Abbreviations: ECG, electrocardiogram; IQR, interquartile range; POAF, postoperative atrial fibrillation.

Figure 2.:
The postoperative day on which atrial fibrillation was first registered.

Episodes of Spo₂ <85% for ≥5 consecutive minutes occurred in 8 of 26 patients before POAF (31%; 95% CI, 14–52) versus 117 of 372 patients (32%; 95% CI, 27–36) without POAF, adjusted OR 1.02 (95% CI, 0.24–4.00; P = .98). RR ≥25·min⁻¹ for ≥5 consecutive minutes occurred in 35% (95% CI, 17–56) vs 17% (95% CI, 13–21) of patients, unadjusted OR 2.59 (95% CI, 0.97–6.48; P = .03) but the adjusted estimate was not statistically significant (Table 1).

The frequency of serious adverse events within 30 days of surgery was 65% (95% CI, 38–86) in the 17 patients with POAF that converted to sinus rhythm spontaneously, 56% (95% CI, 21–86) in the 9 patients with POAF that did not convert spontaneously, and 47% (95% CI, 42–52) in the 372 patients without development of POAF. Hospital length of stay was median 9.4 days (IQR, 6.5–16) in patients with POAF and 6.5 days (IQR, 2.5–11) in patients without POAF (Table 3). The 30-day mortality was 0% (95% CI, 0–15) in patients with POAF versus 0.3% (95% CI, 0.0–1.7) patients without POAF, and 6-month mortality was 12% (95% CI, 2.5–30) in patients with POAF versus 3.5% in (95% CI, 1.9–5.9) patients without POAF.

DISCUSSION

This observational study using wireless close monitoring with repeated sampling of vital signs found a 6.5% frequency of POAF lasting at least 30 consecutive minutes among 398 patients admitted to general surgical wards following major gastrointestinal cancer surgery. Patients with episodes of POAF had 9 days of postoperative hospital admission, but deviations of vital signs including severe hypoxia were not significantly associated with subsequent POAF.

The frequency of POAF (6.5%; 95% CI, 4.5–9.4) in this entirely elective surgical population was high compared with previous studies on patients admitted to general surgical wards (0.4%–4.4%).^{^5–10} Detection of POAF outside the PACU and the intensive care unit relies on manual measurements of vital signs with intervals being up to 12 hours.^¹¹ In the present study, we were able to detect asymptomatic cases of POAF in the periods in between these routine assessments which is a novel approach for detection of vital signs in this group of patients, thus identifying POAF in a subgroup of patients normally not detected by clinical staff.

The etiology and pathophysiology of POAF after noncardiac surgery are not well understood, and vital sign deviations may reveal both precipitating and alerting factors. By close monitoring, we were unable to detect significant associations between hypoxia episodes or other deviations and subsequent POAF. It is of interest that >30% of patients experienced at least 1 episode of Spo₂ <85% for at least 5 minutes. This observation is in line with previous findings of a high incidence of episodes with desaturation after gastrointestinal cancer surgery^¹³ and emphasizes the need for continued close monitoring of postoperative patients after discharge from the PACU. One possible solution is the application of wireless monitoring devices recording vital signs continuously, as also used in this study.

In total, we logged 108 incidents of estimated AF at the specific minute it occurred. Incidents were lasting at least 30 minutes each, but some of these periods were relapses after previously identified AF with a period of regular heart rhythm in-between. As Figure 3 illustrates, patients converted back and forth between regular heart rhythm and shorter periods of AF several times, accentuating the complex challenge of detecting POAF without close monitoring. Recent studies of patients with cardiac implantable electronic devices indicate that as little as 6 minutes of AF might increase the risk of stroke thereby suggesting that anticoagulants can be indicated even in patients experiencing self-limiting POAF of short duration.^¹⁸ Conversely, findings of the recently published LOOP (Implantable Loop Recorder Detection of Atrial Fibrillation to Prevent Stroke) study indicate that patients with as short as 6 minutes of asymptomatic AF do not merit anticoagulant treatment as the associated stroke risk in these patients might be lower.^¹⁹ We report the occurrence of POAF lasting 30 minutes or longer, but it remains to be established which minimum interval and which conditions of AF that are clinically and socioeconomically relevant to start treatment for.

Figure 3.:
Atrial fibrillation after major noncardiac surgery. Duration of monitoring period (gray line) in patients with up to 4 days of single-lead ECG monitoring after PACU discharge, and duration of each incident of postoperative atrial fibrillation (black line). ECG indicates electrocardiogram; PACU, postanesthesia care unit.

By examining the final 5 minutes of ECG recordings for each patient, we observed that 27% of the patients with POAF still had POAF when the monitoring ended, while 65% of the patients reverted to sinus rhythm spontaneously during the monitoring period, and only 8% reverted to sinus rhythm after receiving antiarrhythmic treatment. Previous reports have indicated that patients with POAF commonly revert to sinus rhythm during hospitalization, but in most cases after administration of antiarrhythmic agents.^³^,^⁵ This is important as prompt antiarrhythmic treatment may not be indicated among otherwise stable patients with POAF. Also, device-detected AF in patients with implanted pacemakers suggests that many of the patients with AF at the time of hospitalization will have a chronic or recurring pattern of AF indicating that patients with POAF in the current study are likely to experience AF after hospital discharge.^²⁰

By close monitoring of heart rhythm, we found 26 patients with POAF, while 14 of these patients were discovered by clinical staff during 4 days of postoperative monitoring. Many patients remained hospitalized, but 35% of patients with POAF were still undiscovered by clinical staff at the time of hospital discharge. A recent study of the VISION (Vascular Events in Noncardiac Surgery Cohort Evaluation) cohort argues that shorter periods of asymptomatic POAF are likely to be detected more often with active screening,^¹⁰ which underlines the potential benefit of wireless close heart rhythm monitoring in detecting asymptomatic cases of POAF in general surgical wards where close vital signs monitoring is not standard care.

Hospital length of stay had the highest point estimate in patients with POAF, and this was formally statistically significant with the Wilcoxon rank sums test, although we recognize that this did not adjust for important confounders and no statistically significant difference was found for postoperative mortality. Prolonged length of stay was mainly caused by serious adverse events that followed POAF in 21 of the 26 patients. Comparatively, previous studies have found an increased length of stay (2.9 days versus 11–34 days) and increased mortality during hospital admission in patients with POAF (0% vs 7%–14%).^{^5–7} The discrepancies between the findings of our and previous studies may perhaps be that most cases of POAF in our study were asymptomatic and undetected by clinical staff, which may signify a lower severity of disease.

This study with >30,000 hours of patient monitoring time using wireless close monitoring with repeated sampling of vital signs presents a novel approach to estimating and detecting POAF outside of telemetry units. This allowed for highly sensitive detection of deviations in vital signs as well as onset and remission of POAF in patients admitted to general surgical wards. We detected the occurrence of POAF by evaluation of R-wave peak intervals from single-lead ECG recordings using a novel purpose-built, computerized algorithm. The risk of false-positive classification was minimized by the adjudication of the detected cases by 2 cardiology specialists. POAF may, however, occur at even higher frequencies than found in our study. First, although the algorithm underwent extensive internal validation before application and the detection of vital signs used the same technique as in other studies,^¹³^,^¹⁴^,^¹⁶^,^²¹ there is a risk that not all relevant cases were detected (ie, false negatives). Second, we only monitored patients in the time interval between PACU discharge and a maximum of 4 subsequent admission days. Some patients were discharged from the hospital on the second postoperative day, and some cases of POAF could have presented beyond our follow-up, as well as before start of monitoring due to prolonged stay in PACU, meaning some patients were monitored only from the first postoperative day and onward. Third, when comparing our findings with previous studies, it is important to consider the signal required in each study to define POAF. We required 30 consecutive minutes of AF recorded with a sampling frequency of 10 seconds per minute. Others relied on single or daily ECG measurements or International Classification of Diseases, 10th revision, codes in patients’ charts, and their diagnosis of POAF was confirmed after an unknown or unregistered duration of AF,^³^,^⁵^,^⁶^,^{^8–10}^,^²²^,^²³ or if AF was persistent for >10 minutes.^⁷ Additional patients in this study have likely experienced episodes of AF <30 minutes, but we regarded the duration of AF to be of clinical relevance and with minimal risk of false positives. Although we detected several cases of POAF, it must be emphasized that further studies are needed to clarify optimal thresholds for alerts to clinical staff. The economic cost of this added monitoring, diagnostics, and treatment should clearly be documented against evidence for improved patient outcome.

In conclusion, we found that new-onset POAF occurred in 6.5% (95% CI, 4.5–9.4) of patients after gastrointestinal cancer surgery, which is high compared with previous findings in these patients. Episodes of deviating vital signs were not associated with subsequent development of POAF although this study did not have power to assess this fully.

DISCLOSURES

Name: Johan D. V. Jokinen, MD.

Contribution: This author helped conceive the idea and study design; helped collect, analyze, and interpret data; helped review the literature and write all drafts; and approved the final manuscript.