Postoperative atrial fibrillation in emergent non-cardiac surgery: Risk factors and outcomes from a ten-year intensive-care unit retrospective study

Abstract

BACKGROUND

Atrial fibrillation (AF) represents a common arrhythmia with significant implications and may occur pre-, intra-, or postoperatively (POAF). After cardiac surgery POAF occurs in approximately 30% of patients, while non-cardiac/non-thoracic surgery has a reported incidence between 0.4% to 15%, with new onset POAF occurring at a rate of 0.4% to 3%. While AF has been extensively studied, it has not been well described in emergent non-cardiac surgery associated with increased surgical stress in an intensive care unit setting (ICU).

AIM

To investigate the incidence/predictors of POAF in emergent non-cardiac surgery and its associations with postoperative outcomes in the ICU.

METHODS

This retrospective study included patients ≥ 18 years who underwent exploratory laparotomy or lower extremity amputation between October 2012 and September 2023 and were admitted in the ICU. Data of interest included occurrence of POAF, demographic characteristics, comorbidities, laboratory values, administered fluids, medications, and postoperative outcomes. Statistical analyses consisted of identifying predictors of POAF and associations of POAF with outcomes of interest.

RESULTS

A total of 347 ICU patients were included, 16.4% had a history of AF, 13.0% developed POAF, and 7.9% developed new-onset POAF. Patients with new-onset POAF were older (79.6 ± 9.1 vs 68.1 ± 14.8 years, < 0.001), of white race (47.8% vs 28.8, P < 0.001), hypertensive (87.0% vs 71.2%, P = 0.011), had longer ICU length of stay (ICU-LOS) (13.4 vs 6.7 days, P = 0.042), higher mortality (43.5% vs 17.6%, P = 0.016) and higher rate of cardiac arrest (34.8% vs 14.6%, P = 0.005) compared to patients without new-onset POAF. Multivariable analysis revealed increased POAF risk with advanced age (OR = 1.06; 95%CI: 1.02-1.10, P = 0.005), white race (OR = 2.85; 95%CI: 1.26-6.76, P = 0.014), high intraoperative fluid (OR > 1; 95%CI: 1.00-1.00, P = 0.018), and longer ICU-LOS (OR = 1.04; 95%CI: 1.00-1.08, P = 0.023). After adjusting for demographics, new onset POAF significantly predicted mortality (OR = 3.07; 95%CI: 1.14-8.01, P = 0.022).

CONCLUSION

POAF was associated with prolonged ICU-LOS, white race, and high intraoperative fluid. New-onset POAF was associated with increased risk of cardiac arrest and death in critically ill patients.

Key Words: Postoperative atrial fibrillation; Non-cardiac surgery; Intensive care unit; Postoperative mortality; Emergent surgery; Exploratory laparotomy; Knee amputation

Core Tip: In critically ill patients undergoing emergent non-cardiac surgery, such as exploratory laparotomy or amputation, postoperative atrial fibrillation (POAF) occurred in 13.0% and 7.9% experienced new onset POAF. Factors associated with POAF included advanced age, white race, high intraoperative fluid intake, and prolonged intensive care unit stay. New onset POAF was linked to higher mortality and an increased risk of cardiac arrest. These findings suggest the need for increased awareness of these potential predictors, vigilant monitoring and management in high-risk surgical patients to prevent POAF and associated complications.

INTRODUCTION

Atrial fibrillation (AF) is the most common complication after cardiac surgery occurring in approximately 30% of this population and the most common type of secondary AF[1,2]. Postoperative AF (POAF) (with or without prior AF history) also affects patients undergoing non-cardiac/non-thoracic surgery, with an incidence between 0.4% to 15% and is associated with significant morbidity and mortality[2-4]. The incidence of new-onset POAF (patient with no history of AF developing POAF) after non-cardiac surgery has been estimated between 0.4% to 3%[3,5,6].

The pathogenesis of POAF is multifactorial, involving mechanisms such as inflammation, oxidative stress, autonomic dysregulation, release of catecholamines, and direct myocardial injury/ischemia[2,7]. Preoperative factors such as advanced age, male gender, history of AF, and comorbid cardiovascular diseases have been shown to increase the risk of POAF[8-10]. Intraoperative and postoperative factors including anemia, hypotension, electrolyte and metabolic disturbances, changes in the intravascular volume (most commonly hypervolemia), and pain contribute to the development of POAF[8-12].

The clinical significance of POAF extends beyond transient arrhythmia, as it is associated with increased risk of thromboembolic events, heart failure, and prolonged hospital stay[13-15]. POAF that occurs within the first postoperative month after non-cardiac surgery has been shown to recur at subsequent follow-up at 1-year (18%) and 2-years (16.8%) after the index operation and is associated with increased risk of thromboembolism, major bleeding, and mortality[1,16,17]. POAF in non-cardiac surgery has been associated with at least 4-fold increased risk of stroke and at least 3-fold higher risk of mortality in the long-term[18].

POAF has been extensively studied in the context of cardiac surgery, but there is limited data on its occurrence in emergent non-cardiac surgery. This study aims to identify risk factors for POAF in emergent non-cardiac surgeries and to examine the relationships between these risk factors in patients with and without a prior history of AF. Additionally, we assessed the associations between POAF and relevant clinical outcomes.

MATERIALS AND METHODS

Setting and participants

This retrospective study included patients aged 18 years or older who underwent either emergent exploratory laparotomy or lower extremity amputation (above the knee (AKA) or below the knee amputation (BKA) between October 2012 and September 2023 and were admitted in the intensive care unit (ICU) at a community hospital setting in the New York area. Emergent exploratory laparotomy in this study refers to a surgical procedure that was performed to investigate the cause of acute abdominal symptoms when the diagnosis was unclear in emergency situations, including peritonitis due to abdominal organ perforation, intraabdominal abscess or other causes of intra-abdominal sepsis, bowel obstruction, uncontrolled gastrointestinal bleeding, or blunt or penetrating trauma. Institutional IRB approval was granted with study number 2107699-1 (letter attached). Epic System Inc.'s Slicer Dicer application was used to extract the records of eligible patients and variables of interest were automatically or manually extracted by five authors.

Study design

The study population was grouped into four groups: Patients with new-onset POAF and no prior history of AF ("New onset POAF"), patients with POAF and a known history of AF (POAF, hx), patients without POAF but with a history of AF (No POAF, hx), and patients without POAF or history of AF (No POAF, no hx).

Variables

Variables of interest included demographic characteristics [age, gender, race, body mass index (BMI)], important dates [date of admission, date of procedure, date of discharge, ICU length of stay (ICU-LOS)], comorbidities [history of AF, hypertension (HTN), diabetes mellitus (DM), hyperlipidemia (HLD), coronary artery disease (CAD), congestive heart failure (CHF), chronic kidney disease (CKD), previous history of stroke or transient ischemic attack (TIA), cancer, smoking status], procedure related data [type of procedure (AKA, BKA, exploratory laparotomy), duration of procedure, type of anesthesia, and intraoperative IV fluid (IVF) intake captured via manual review of intraoperative anesthesiologist’s records], preoperative b-blockers, echocardiography parameters [left ventricular ejection fraction (LVEF), left atrium (LA) diameter], key laboratory parameters [hemoglobin (Hgb), sodium (Na), magnesium (Mg), potassium (K), phosphorus (P), calcium (Ca), brain natriuretic peptide (BNP), lactate], main outcomes/complications [POAF, new onset POAF, stroke, myocardial infarction (MI), major adverse limb event, systemic embolism, cardiac arrest, major bleeding according to ISTH criteria[19], all-cause mortality] and treatment of POAF [beta-blocker, calcium channel blocker (CCB), amiodarone, electrical cardioversion, direct oral anticoagulant (DOAC), warfarin, heparin infusion].

Analytic plan

Bivariate analyses were conducted to identify significant differences among the POAF groups. Multivariable models were used to examine predictors of POAF and in-hospital mortality. Continuous variables were described with mean and SD or medians and interquartile ranges (IQR), depending on the shape of the variable distribution. Categorical variables were described using percentages and frequencies. Analysis of variance (ANOVA) and χ² tests, or their non-parametric equivalents, were used to test differences in POAF groups in continuous and categorical variables, respectively. Variables found to have significant differences in bivariate analysis were examined using post-hoc analysis adjusting for multiple comparisons.

Characteristics used in logistic regression models were based on these bivariate comparisons. Selection for inclusion into multivariable analysis predicting POAF and in-hospital mortality was determined by relations identified in bivariate and post-hoc testing that showed significant differences between patients in the new onset POAF group compared to other groups. Variables selected for inclusion in final models were assessed for levels of missingness. Variables with levels of missingness above established guidelines[20], were compared across the primary independent variable (AF group), selected covariates, and primary outcomes via sensitivity analysis, including both bivariate and multivariable analysis. Complete data was available for both the primary independent variable and primary outcomes.

To understand both relations between POAF groups and selected covariates to in-hospital mortality, predictors were added to the model in blocks; first unadjusted OR were estimated for POAF groups, then demographic characteristics were added, followed by procedural characteristics and ICU-LOS. Given that there is limited literature on the risk factors of POAF and new onset POAF, the predictive models were considered exploratory. As a result, the covariates selected for inclusion into multivariable models were determined by examining those which were related to AF in bivariate analyses. Specifically, covariates which were shown to significantly differ between new onset AFIB and other groups were selected for the final models and all characteristics were added in a single block. Patients without POAF or history of AF were used as the reference group for both models. A value of P < 0.05 was considered statistically significant. All statistical analyses were performed with the use of R language version 4.2.3 within the RStudio software version 2023.03.0+386.

RESULTS

Descriptive statistics

Demographic, clinical, and laboratory characteristics: The study sample comprised 347 patients, with a mean age of 70.2 ± 14.3 years (Figure 1; Table 1). The sample consisted of 49.6% (n = 172) females and 50.4% (n = 175) males. Patients identified their race primarily as White (34.9%, n = 121), Black/African-American (20.5%, n = 71), or Hispanic/Latino (19.3%, n = 67). The BMI of the sample was 27.0 ± 7.3 kg/m².

Figure 1 Population selection process and atrial fibrillation groups. POAF: Postoperative atrial fibrillation; ICU: Intensive care unit; hx: History.

Table 1 Baseline characteristics of the study population, n (%).

Variables of interest	POAF, hx (n = 22)	POAF, no hx (n = 23)	No POAF, hx (n = 35)	No POAF, no hx (n = 267)	Overall (n = 347)	P value
Demographics
Age, mean (SD)	76.7 (9.4)	79.6 (9.1)	76.1 (10.7)	68.1 (14.8)	70.2 (14.3)	< 0.001
Male	14 (63.6)	10 (43.5)	21 (60)	130 (48.7)	175 (50.4)	0.316
BMI, mean, (SD)	29.3 (8.7)	27.1 (4.5)	26.9 (6.9)	26.8 (7.4)	27.0 (7.3)	0.405
Race						0.001
Asian	3 (13.6)	4 (17.4)	5 (14.3)	51 (19.1)	63 (18.2)
Black	2 (9.1)	5 (21.7)	6 (17.1)	58 (21.7)	71 (20.5)
Hispanic	0 (0.0)	1 (4.3)	3 (8.6)	63 (23.6)	67 (19.3)
White	16 (72.7)	11 (47.8)	17 (48.6)	77 (28.8)	121 (34.9)
Comorbidities
Hypertension	19 (86.4)	20 (87.0)	32 (91.4)	190 (71.2)	261 (75.2)	0.014
Diabetes mellitus	9 (40.9)	8 (34.8)	16 (45.7)	125 (46.8)	158 (45.5)	0.706
Congestive heart failure	4 (18.2)	2 (8.7)	12 (34.3)	31 (11.6)	49 (14.1)	0.006
Stroke/transient ischemic attack history	2 (9.1)	2 (8.7)	13 (37.1)	34 (12.7)	51 (14.7)	0.004
Coronary artery disease	8 (36.4)	3 (13.0)	12 (34.3)	44 (16.5)	67 (19.3)	0.013
Hyperlipidemia	9 (40.9)	7 (30.4)	16 (45.7)	78 (29.2)	110 (31.7)	0.179
Chronic kidney disease	4 (18.2)	1 (4.3)	6 (17.1)	39 (14.6)	50 (14.4)	0.483
Cancer	2 (9.1)	6 (26.1)	5 (14.3)	62 (23.2)	75 (21.6)	0.296

POAF: Postoperative atrial fibrillation; hx: History; BMI: Body mass index.

Comorbidities included HTN in 261 patients (75.2%), DM in 158 patients (45.5%), HLD in 110 patients (31.7%), CAD in 67 patients (19.3%), CHF in 49 patients (14.1%), CKD in 50 patients (14.4%), a previous history of stroke or TIA in 51 patients (14.7%), a history of cancer in 75 patients (21.6%), while 103 patients (29.7%) were active smokers.

The mean Hgb was 8.5 ± 1.7 g/dL, serum Na was 135 ± 5.1 mEq/L, serum Mg was 1.6 ± 0.3 mg/dL, median serum K was 3.6 mEq/L (IQR = 3.2-3.8), median P was 2.7 mg/dL (IQR = 2.2-3.5), and median Ca 7.3 mg/dL (IQR = 6.7-8.0) (Table 2). The median serum BNP was markedly elevated (3270 pg/mL, IQR = 1180-10000). The median lactate was 1.9 mmol/L (IQR = 1.2-3.5). Echocardiography revealed a mean LA diameter of 39.5 ± 6.0 mm. The mean LVEF was 56.9% ± 10.2%.

Table 2 Main laboratory parameters, n (%).

Variables of interest	POAF, hx (n = 22)	POAF, no hx (n = 23)	No POAF, hx (n = 35)	No POAF, no hx (n = 267)	Overall (n = 347)	P value
Laboratory parameters (mean/SD, median/IQR)
Hemoglobin (mean, SD)	8.7 (1.7)	8.4 (1.8)	8.0 (1.4)	8.5 (1.8)	8.5 (1.7)	0.344
Brain natriuretic peptide (median, IQR)	5090 (3770-9840)	18900 (8740-29500)	2590 (1100-5960)	2420 (918-7810)	3270 (1180-10000)	0.002
Lactate (median, IQR)	1.5 (1.3-2.3)	3.3 (2.2-5.3)	2.3 (1.5-3.1)	1.8 (1.2-3.3)	1.9 (1.2-3.5)	0.091
Left Atrium diameter (mean, SD)	43.6 (5)	38.7 (3.5)	43.6 (7.7)	38.5 (5.6)	39.5 (6)	< 0.001
Left ventricular ejection fraction (mean %, SD)	57.0 (8.5)	57.7 (9.3)	55.3 (10.4)	57.0 (10.5)	56.9 (10.2)	0.692
Electrolytes
Sodium (mean, SD)	136 (4.1)	133 (7.2)	136 (5.8)	134 (4.8)	135 (5.1)	0.186
Potassium (median, IQR)	3.5 (3.1-3.8)	3.3 (3.1-4)	3.8 (3.5-4.2)	3.5 (3.2-3.8)	3.6 (3.2-3.8)	0.130
Calcium (median, IQR)	7.1(6.5-7.6)	7 (6.5-7.5)	7.5 (7-8.3)	7.4 (6.7-8.0)	7.3 (6.7-8.0)	0.076
Magnesium (mean, SD)	1.7 (0.3)	1.6 (0.3)	1.7 (0.3)	1.6 (0.3)	1.6 (0.3)	0.688
Phosphorus (median, IQR)	2.7 (2.1-3.2)	2.5 (2.2-3.9)	2.8 (2.2-3.8)	2.7 (2.2-3.5)	2.7 (2.2-3.5)	0.977

POAF: Postoperative atrial fibrillation; hx: History; IQR: Interquartile range.

Procedural characteristics: In terms of non-cardiac procedures selected, 76.9% (n = 267) of the population underwent an exploratory laparotomy, 16.1% (n = 56) had an AKA, and 6.6% (n = 23) had a BKA (Table 3). The mean duration of surgery was 2.6 ± 1.2 hours. General endotracheal anesthesia was utilized in 279 patients (80.4%), regional anesthesia in 36 patients (10.4%), combination of anesthesia techniques in 28 patients (8.1%) and monitored anesthesia care in 4 patients (1.2%). A total of 87 patients were on preoperative beta-blocker therapy (25.1%). Intraoperative fluid intake was on average 2240 ± 1920 cc.

Table 3 Procedural characteristics, n (%).

Variables of interest	POAF, hx (n = 22)	POAF, no hx (n = 23)	No POAF, hx (n = 35)	No POAF, no hx (n = 267)	Overall (n = 347)	P value
Procedural characteristics
Type of procedure
Above the knee amputation	2 (9.1)	1 (4.3)	15 (42.9)	38 (14.2)	56 (16.1)	< 0.001
Below the knee amputation	0 (0.0)	0 (0.0)	3 (8.6)	20 (7.5)	23 (6.6)	0.388
Exploratory laparotomy	20 (90.9)	22 (95.7)	17 (48.6)	208 (77.9)	267 (76.9)	< 0.001
Duration of procedure, [mean, (SD)]	2.9 (1.3)	2.7 (1.0)	2.3 (1.1)	2.6 (1.2)	2.6 (1.2)	0.169
Preoperative b-blocker	7 (31.8)	7 (30.4)	12 (34.3)	61 (22.8)	87 (25.1)	0.322
Type of anesthesia
general endotracheal anesthesia	20 (90.9)	17 (73.9)	25 (71.4)	217 (81.3)	279 (80.4)	0.0764
Regional	1 (4.5)	1 (4.3)	8 (22.9)	26 (9.7)	36 (10.4)
Monitored anesthesia care	0 (0.0)	1 (4.3)	1 (2.9)	2 (0.7)	4 (1.2)
Combined anesthesia techniques	1 (4.5)	4 (17.4)	1 (2.9)	22 (8.2)	28 (8.1)
Mean intraoperative IVF intake [mean, (SD) mL]	2900 (2040)	2420 (1340)	1340 (1420)	2280 (1980)	2240 (1920)	0.001

POAF: Postoperative atrial fibrillation; hx: History; IVF: Intravenous fluid.

Outcomes: In a total of 347 patients, 16.4% (n = 57/347) had a history of AF, 13.0% (n = 45/347) developed POAF, and 7.9% (n = 23/290; where n = 290 is the population without AF history and at risk of new onset POAF) had new-onset POAF (Table 4).

Table 4 Main outcomes and atrial fibrillation treatment, n (%).

Variables of Interest	POAF, hx (n = 22)	POAF, no hx (n = 23)	No POAF, hx (n = 35)	No POAF, no hx (n = 267)	Overall (n = 347)	P value
Outcome
Stroke	0 (0.0)	0 (0.0)	1 (2.9)	8 (3.0)	9 (2.6)	1.000
Myocardial infarction	2 (9.1)	3 (13.0)	0 (0.0)	6 (2.2)	11 (3.2)	0.018
Major adverse limb event	1 (4.5)	0 (0.0)	0 (0.0)	0 (0.0)	1 (0.3)	0.064
Systemic embolism	0 (0.0)	0 (0.0)	0 (0.0)	6 (2.2)	6 (1.7)	1.000
Major bleeding	7 (31.8)	6 (26.1)	5 (14.3)	38 (14.2)	56 (16.1)	0.093
Cardiac arrest	7 (31.8)	8 (34.8)	2 (5.7)	39 (14.6)	56 (16.1)	0.005
Postoperative day of POAF (mean, SD days)	8.9 (19.4)	7.9 (7.6)	-	-	-	0.073
ICU-LOS in days (median, IQR)	11.5 (6.3-17.8)	9.0 (5.0-22.0)	5.0 (0.0-11.5)	4.0 (1.0-8.0)	5.0 (2.0-10.0)	< 0.001
All-cause mortality	6 (27.3)	10 (43.5)	3 (8.6)	47 (17.6)	66 (19.0)	< 0.001
POAF treatment
B-blocker	16 (72.7)	13 (56.5)	-	-	-	0.353
Calcium channel blockers	5 (22.7)	5 (21.7)	-	-	-	1.000
Amiodarone	6 (27.3)	6 (26.1)	-	-	-	1.000
DOAC	3 (13.6)	1 (4.3)	-	-	-	0.346
Warfarin	0 (0.0)	2 (8.7)	-	-	-	0.489
Heparin infusion	1 (4.5)	2 (8.7)	-	-	-	1.000

POAF: Postoperative atrial fibrillation; hx: History; IQR: Interquartile range; IVF: Intravenous fluid; ICU: Intensive care unit; LOS: Length of stay; DOAC: Direct oral anticoagulant.

Among patients with POAF (n = 45), the most frequently utilized treatment modalities included beta-blockers (64.4%, n = 29/45), non-dihydropyridine CCBs (22.2%, n = 10/45), and amiodarone (26.7%, n = 12/45). No patients underwent cardioversion.

The median ICU-LOS was 5 days (IQR = 2-10). Postoperative complications included postoperative major bleeding in 16.1% (n = 56/347), fatal bleeding in 1.2% (n = 4), cardiac arrest in 16.1% (n = 56/347), MI in 3.2% (n = 11/347), major adverse limb event in 0.3% (n = 1/347) and systemic embolism in 1.7% (n = 6/347). In the new onset POAF group, postoperative MI occurred in 13.0% patients (n = 3/23) and cardiac arrest in 34.8% (n = 8/23). Across all patients, the mortality was 19.0% (n = 66) and occurred at a median of 10 days (IQR = 4-21).

Bivariate analysis

Demographic and clinical characteristics: Compared to other groups, patients with new onset POAF were significantly older (mean age 79.6 ± 9.1 years) and were more likely to be White (n = 16/22, 72.7%) (P < 0.01) (Table 1). HTN was less common in patients without POAF and no history of AF (P = 0.014), while CHF and a history of stroke or TIA were more common in those with a history of AF but no POAF (P = 0.002 and P = 0.0006, respectively).

Procedural characteristics: Compared to patients without AF history, patients with prior AF had significantly larger mean LA diameter (38.6 mm vs 43.6 mm, respectively P = 0.03) (Table 3). Among patients with a history of AF, the IVF intake was significantly higher in patients with POAF compared to those without POAF (2900 mL vs 1340 mL, P = 0.001). Among patients without POAF, IVF were administered more deliberately in patients without history of AF compared to those with a prior history (2280 mL vs 1340 mL, P = 0.044).

Outcomes: Overall, cardiac arrest was significantly more common in patients with new onset POAF (34.8%, n = 8/23) compared to patients without POAF and no history of AF (14.6%, n = 39/267) (P = 0.005) (Table 4). Although not statistically significant, patients with POAF (with or without AF history) had the highest rates of postoperative major bleeding across groups 31.8% with history of AF vs 26.1% without history of AF compared to 14.3% in patients without POAF (P = 0.0926). Patients with new onset POAF had a longer median ICU-LOS (9 days, IQR = 5-22) compared to those without POAF and no history of AF (4 days, IQR = 1-8) (P = 0.042). Mortality was significantly higher in the new onset POAF group compared to all other groups (43.5%, P = 0.016). Among patients who expired, postoperative day of death occurred after the longest interval among new onset POAF (median 15 days, IQR = 8.5-27.5).

Multivariable analysis

POAF occurrence: In multivariable logistic regression, new-onset POAF was significantly associated with advanced age (OR = 1.08; 95%CI: 1.02-1.15, P = 0.008) (Figure 2A; Table 5). However, other factors including identifying as white, LA diameter, intraoperative IVF intake, and ICU-LOS were not associated with new onset POAF.

Figure 2 Multivariable analyses. A: New-onset postoperative atrial fibrillation; B: Postoperative (with or without atrial fibrillation history) atrial fibrillation; C: In-hospital mortality. POAF: Postoperative atrial fibrillation; ICU: Intensive care unit; LOS: Length of stay; hx: History.

Table 5 Multivariable analyses: Postoperative atrial fibrillation.

Variables of interest	New onset POAF (without AF hx)			POAF (with/without AF hx)
Variables	OR	95%CI	P value	OR	95%CI	P value
Age	1.08	1.02-1.15	0.008	1.06	1.02-1.10	0.005
Race: White	1.48	0.48-4.62	0.493	2.85	1.26-6.76	0.014
Left atrium diameter	0.96	0.86-1.06	0.474	1.06	0.99-1.13	0.076
Intraoperative IVF intake	1	1.00-1.00	0.284	1	1.00-1.00	0.018
ICU-LOS	1.02	0.97-1.07	0.336	1.04	1.00-1.08	0.023

POAF: Postoperative atrial fibrillation; AF: Atrial fibrillation; hx: History; IVF: Intravenous fluid; ICU: Intensive care unit; LOS: Length of stay.

Occurrence of POAF, with or without a history of AF, was significantly associated with advanced age (OR = 1.06, 95%CI: 1.02-1.10, P = 0.005), identifying as white (OR = 2.85, 95%CI: 1.26-6.76, P = 0.014), higher intraoperative IVF intake (OR = 1.00; 95%CI: 1.00-1.00, P = 0.018), and prolonged ICU-LOS (OR = 1.04, 95%CI: 1.00-1.08, P = 0.023) (Figure 2B).

In-hospital mortality: In the multivariable analysis of in-hospital mortality, the first model, which included only the POAF groups, showed that patients with new-onset POAF had 3.6-fold higher odds of death compared to those without POAF and no history of AF (No POAF, no hx) (OR = 3.60; 95%CI: 1.46-8.69, P = 0.004) (Figure 2C; Table 6) In the second model, after adjusting for demographic characteristics such as age and race, new onset POAF remained a significant predictor of mortality (OR = 3.07, 95%CI: 1.14-8.01, P = 0.022). However, in the third model, which further adjusted for LA diameter, IVF intake, and ICU-LOS, only prolonged ICU-LOS remained significantly associated with in-hospital mortality (OR = 1.05, 95%CI: 1.02-1.09, P = 0.002).

Table 6 Multivariable analyses: Mortality.

	Model 1			Model 2			Model 3
Variables	OR	95%CI	P value	OR	95%CI	P value	OR	95%CI	P value
POAF, hx	1.76	0.60-4.52	0.265	1.51	0.49-4.13	0.444	0.75	0.19-2.59	0.667
POAF, no hx	3.60	1.46-8.69	0.004	3.07	1.14-8.01	0.022	0.82	0.19-2.84	0.773
No POAF, hx	0.44	0.10-1.29	0.187	0.44	0.10-1.35	0.201	0.23	0.03-1.01	0.091
Age	-	-	-	1.01	0.99-1.03	0.408	1.02	0.99-1.05	0.248
Race: White	-	-	-	1.58	0.86-2.89	0.134	1.26	0.61-2.59	0.532
Left atrium diameter	-	-	-	-	-	-	1.00	0.93-1.06	0.908
Intraoperative IVF intake	-	-	-	-	-	-	1.00	1.00-1.00	0.096
ICU-LOS	-	-	-	-	-	-	1.05	1.02-1.09	0.002

POAF: Postoperative atrial fibrillation; hx: History; IVF: Intravenous fluid; ICU-LOS: Intensive care unit; LOS: Length of stay.

DISCUSSION

The mechanisms underlying the development of POAF remain inadequately understood. Understanding the associated risk factors is essential for adequate prevention and management. POAF occurs in 30%-40% of thoracic and cardiac surgeries, and at a lower incidence of 0.4%-3% presents as new onset POAF in non-cardiothoracic surgeries[3,4,6]. The incidence of new onset POAF varies among different types of non-cardiac surgery. Pooled data from a meta-analysis by McIntyre et al[21] published in 2021, that included 346 studies with more than 5 million patients, reported a 13.4% incidence of new onset POAF in patients undergoing esophagectomy (n = 13421), a 5.4% incidence of new onset POAF in patients undergoing vascular surgery (n = 29135), and a 3.3% incidence of new onset POAF in patients undergoing abdominal surgery (n = 194498). Our findings suggest that non-cardiac procedures with high perioperative stress, such as exploratory laparotomy and lower extremity amputation, can trigger POAF in 13.0% of the population and lead to developing new onset POAF in 7.9% of the population at risk, which could be attributed to the emergent procedure and overall status of critically ill patients admitted in the ICU.

AF is characterized by rapid and ineffective atrial depolarization that may result in an irregular ventricular response and poses significant clinical challenges. In the postoperative setting it typically presents within the first four days following non-cardiac surgery[8,22]. The underlying pathophysiology of POAF is driven by several mechanisms, including the activation of the sympathetic nervous system due to factors like pain, hypovolemia, and hemodynamic instability, alongside electrolyte imbalances such as hypokalemia and hypomagnesemia. Additional contributing factors include hypoglycemia, hypoxia, hypervolemia due to aggressive fluid resuscitation, and overt inflammatory response. Predictive factors for POAF include advanced age, male gender, a history of cardiovascular disease (such as CHF, peripheral arterial disease, prior history of arrhythmia), and extensive surgery[8,11,12,23]. This multifaceted etiology highlights the complexity of POAF and emphasizes the importance of identifying at risk patients and addressing these triggers.

POAF has several detrimental effects on cardiopulmonary hemodynamics, particularly when the ventricular rate is not controlled. The loss of atrial contraction (atrial kick) results in decreased diastolic filling time, leading to a reduction in stroke volume and cardiac output. This is particularly critical in patients with compromised ventricular function or those undergoing surgery with significant intravascular and extravascular volume changes[24,25]. Further, the irregular rapid ventricular rate increases the myocardial oxygen demand and results in myocardial ischemia, particularly in individuals with underlying CAD. These hemodynamic disturbances can lead to complications such as hypotension, acute exacerbation of CHF, MI, stroke, systemic embolism, and death[2].

POAF can be paroxysmal or persistent, asymptomatic or with symptoms, and has been associated with higher risk of AF recurrence[15,21,26,27]. POAF symptoms are diverse and often non-specific and may include palpitations, fatigue, shortness of breath, chest pain, hypotension, dizziness. In patients without symptoms, complications such as stroke or systemic embolism may be the first manifestation, especially in patients who are not receiving treatment dose anticoagulation. The diagnosis of AF in the perioperative setting relies heavily on electrocardiographic monitoring. AF is characterized by an irregular rhythm with no discernible P waves on the electrocardiogram, reflecting the disorganized atrial depolarization. Transthoracic echocardiography (TTE) is useful for assessing atrial size, ventricular function, and the presence of structural heart disease. Parameters such as LA volume indexed for body surface, peak atrial longitudinal strain, and total atrial conduction time have been shown as valuable predictors of POAF[28].

Prevention strategies for POAF focus on minimizing perioperative stressors such as electrolyte imbalances, hypovolemia/hypotension and inadequate pain management[29]. Preoperative use of amiodarone, sotalol, or colchicine has been shown to reduce the risk of POAF in both cardiac and non-cardiac surgeries. However, the benefit of these agents may be limited by their adverse effects including hypotension and bradycardia[30-34].

Several studies have attempted to predict POAF in cardiac and non-cardiac surgery[23,35-39]. Most scores were initially developed for different purposes and were predominantly evaluated for the prediction of POAF in cardiac surgery. In a pooled analysis by Pandey et al[39] those scores (including the CHA2DS2-VASc, POAF score, EuroScore and EuroSCORE II, among others) were shown to have moderate/high sensitivity (46%-87%) and low/moderate specificity (31%-70%). In non-cardiac thoracic surgery, the prediction rule by Passman et al[36], which factors in age, gender, and preoperative heart rate as independent predictors based on data from a cohort of 856 patients (147 with POAF), estimates the risk of POAF from 0% (no risk factors) to 54.6% (cumulative score of 6 points). Recently, Oh et al[37] used machine learning techniques in a cohort of 201,864 patients undergoing non-cardiac surgery (5,735 with POAF), and identified five variables including age, lung procedure, duration of the procedure, history of CAD, and HTN as predictors of POAF in a model with area under the curve 0.80 and an accuracy of 0.95. In this study, similar to other studies, both the univariate and multivariable analyses revealed that advanced age remained a significant predictor of new onset POAF, after adjusting for demographics. Advanced age is associated with a reduction in myocardial fibers, increased fibrosis, and greater collagen deposition within the atria, especially around the sinoatrial node, leading to changes in atrial electrical activity[40,41]. These age-related changes create a predisposition for POAF, while the acute stress of surgery and the resulting inflammation likely serve as the triggering factors for its onset[42,43]. We also showed that new onset POAF occurs more frequently in patients identifying as white, and those with a history of HTN. The reasons behind the increased frequency of new onset POAF in patients identifying as white are incompletely understood, but could be attributed to differences in access to healthcare, genetic predisposition or anatomic differences relevant to development of POAF, such as differences in preoperative left atrial sizes with less left atrial enlargement in response to HTN in African Americans[44,45]. However, in this study there was no statistically significant difference between type of anesthesia, duration of procedure or type of procedure in terms of POAF or new onset POAF occurrence.

Elevated BNP has been previously investigated as a predictor of POAF in cardiac and non-cardiothoracic surgery[46-52]. Data from a 2004 meta-analysis by Wazni et al[48] had demonstrated that elevated preoperative plasma BNP was an independent predictor of POAF. A recently published updated systematic review of 20 studies involving 9079 patients showed significantly elevated BNP and NT-proBNP in patients that developed POAF after cardiac surgery[46]. In the non-cardiothoracic setting, BNP has been associated with cardiovascular events (OR = 17.8, 95%CI: 6.56-48.6), including AF, and 30-day mortality (OR = 10.8, 95%CI: 2.29-51.37), with sensitivity of 91% and 84% and specificity of 75% and 66% at cut-offs of 143 pg/mL and 164 pg/mL, respectively[52]. Pooled data from a systematic review and meta-analysis have associated BNP with a composite of cardiovascular complications, including cardiac death, nonfatal MI and AF within 30 days after non-cardiac surgery[50]. Our patients with new onset POAF had higher median BNP level compared to the patients without POAF and no history of AF (18900 vs 2420 pg/mL) even though this was not statistically significant (P = 0.91). Our findings are in accordance with Cuthbertson et al[51] who reported that preoperative BNP levels are higher in patients developing new onset POAF undergoing non-cardiac surgery.

The management of POAF involves a multifaceted approach consisting of identifying and addressing underlying triggers, rate or rhythm control strategies, and anticoagulation therapy in select patients[53]. The management of POAF often involves a choice between rate control and rhythm control strategies and there is adequate data to support rate control as the first step. The POISE trial offered critical insights into the use of perioperative beta-blockers, revealing both potential benefits and significant risks. The study included 8351 patients undergoing non-cardiac surgery and demonstrated that metoprolol reduced the incidence of the composite endpoint of cardiovascular mortality and non-fatal MI/cardiac arrest compared to placebo (5.8% vs 6.9%, respectively, P = 0.0399), at the cost of increased rates of stroke (1.0% vs 0.5%, respectively, P = 0.0053) and overall mortality (3.1% vs 2.3%, respectively, P = 0.0317). These findings raised important concerns regarding the generalizability of perioperative beta-blocker use, particularly the administration of high doses in the absence of tailored protocols for individual patient risk factors. Fewer individuals in the metoprolol group had cardiac revascularization or developed new clinically significant AF than did those in the placebo group, but more receiving metoprolol had clinically significant hypotension and bradycardia that may have contributed to the strokes that often resulted in severe disability or death[54]. The high incidence of perioperative hypotension observed in the metoprolol group likely contributed to these unfavorable outcomes, emphasizing the need for cautious beta-blocker use in the postoperative setting in beta-blocker naïve patients. The effect of beta-blockers in the perioperative period was further investigated in an updated Cochrane systematic review by Blessberger et al[55] in 2019, who included 83 randomized control trials with 14967 patients undergoing non-cardiac surgery. In that study, there was no evidence of any effect of beta-blockers in cerebrovascular events or in the incidence of ventricular arrhythmias, but it was shown that beta-blockers may have a beneficial effect in POAF/flutter (RR = 0.41, 95%CI: 0.21-0.79) and MI (RR = 0.72, 95%CI: 0.60-0.87) at the cost of increased risk of bradycardia (RR = 2.49, 95%CI: 1.74-3.56) and hypotension (RR = 1.40, 95%CI: 1.29-1.51)[55]. The AFFIRM trial, which compared rhythm-control and rate-control strategies in 4060 high-risk patients (older than 65 years or with risk factors for stroke or death), demonstrated no survival advantage with either strategy and showed a non-significant trend of higher 5-year overall mortality in the rhythm control group (23.8 vs 21.3%, P = 0.08) and more adverse drug-relate effects[56]. In the present study, 87 patients (25.1%) were on preoperative beta-blockers and the treatment of POAF included beta-blockers in 64.4% (n = 29/45), non-dihydropyridine CCBs in 22.2% (n = 10/45), and amiodarone in 26.7% (n = 12/45). Current guidelines favor rate control in the management of POAF, mostly based on data from studies including patients undergoing cardiac surgery[15,57-60]. Especially in patients who are hemodynamically stable. Beta-blockers are first-line therapy, followed by non-dihydropyridine CCB for rate control[30]. In patients with hemodynamic instability, synchronized direct current cardioversion may be indicated[53]. Recently, non-pharmacologic methods such as autonomic neuromodulation (vagus nerve stimulation, baroreceptor stimulation among other methods) have been applied in patients undergoing cardiac surgery with very promising results based on data from a meta-analysis showing reduction of POAF incidence by 63%[61].

One of the most critical aspects of managing POAF is balancing the thromboembolic risk against the risk of postoperative bleeding. The CHA2DS2-VASc score is commonly used to assess stroke risk, with scores ≥ 2 indicating a high risk of thromboembolic events. The use of anticoagulation in POAF is controversial and results from ongoing clinical trials are much awaited, such as the ASPIRE-AF trial (NCT03968393) that will investigate the use of DOAC compared to no anticoagulation on major thromboembolism, including stroke, MI, peripheral arterial thrombosis, and symptomatic VTE, and vascular mortality in patients undergoing non-cardiac surgery. A study from Denmark that included 6048 patients that developed new onset POAF after non-cardiac surgery between 1996-2015 showed that the anticoagulation at follow-up decreased the risk of thromboembolism by 48% (HR = 0.52; 95%CI: 0.40-0.67)[6]. In contrast, a retrospective study from Canada followed-up 22007 patients with new onset POAF after non-cardiac surgery (42% with a CHA2DS2-VASc score of greater than or equal to 4) at a mean of 4 years postoperatively and showed no benefit with anticoagulation and a higher risk of major bleeding[62]. Similarly, a study that used data from the Society of Thoracic Surgeons Adult Cardiac Surgery Database, identified 38936 patients after cardiac surgery, including 9861 discharged on oral anticoagulation, and showed that anticoagulation after new onset POAF post-CABG was associated with increased risk of death (HR = 1.16; 95%CI: 1.06-1.26)[63]. Both the ESC guidelines and the ACC/AHA guidelines suggest that long-term anticoagulation is reasonable to prevent stroke in patients with POAF after cardiac and non-cardiac surgery, when deemed safe in terms of postoperative bleeding[15,64].

The HAS-BLED score (HTN, abnormal renal/Liver function, stroke, bleeding, labile international normalized ratio, elderly, drugs/alcohol) is used to estimate bleeding risk, and has been shown to also predict cardiovascular events (stroke, acute coronary syndrome, acute heart failure) and mortality in patients with AF[65]. In the present study, anticoagulation for POAF included DOAC in 8.9% (n = 4/45), warfarin in 4.4% (n = 2/45), and heparin infusion in 6.7% (n = 3/45). Postoperative major bleeding occurred in 16.1%, with a higher rate in patients with POAF (28.9%) compared to 14.3% in patients without POAF (P = 0.0926). Patients with new onset POAF should follow-up as outpatient to determine continuation of anticoagulation, given the high risk of AF recurrence[15].

The prognosis of patients with POAF depends on the timely identification and management of the arrhythmia. While most episodes are brief, self-resolving and patients spontaneously revert to sinus rhythm[66], one-third of patients with new onset POAF after non-cardiac surgery or medical illness have recurrent AF within 1 year[67]. POAF in non-cardiac surgery has been associated with at least 4-fold increased risk of stroke and at least 3-fold higher risk of mortality in the long-term[18]. Further, POAF is associated with increased short- and long-term risks, including stroke/TIA, MI, subsequent arrhythmias, heart failure, and mortality[1,16,68]. Our patients with new-onset POAF had a similar risk of death (3.6-fold higher odds), but there were no stroke cases identified in patients with new-onset POAF or POAF, and the overall rate in the population that was studied was 2.6%.

Strengths and limitations

This study addresses a gap in the literature by focusing on the occurrence and impact of POAF in emergent non-cardiac surgery in an ICU setting, a population with increased surgical stress that has not been well studied. Our findings are in accordance with previous studies in patients with non-cardiac surgery. However, this study also has limitations that should be considered when interpreting the results. First, the retrospective design introduces potential selection bias and limits the ability to establish correlation. Second, the results from a single-center study of patients with selected emergent procedures may not be generalizable to other healthcare settings with different patient populations and procedures. The procedures that we included affect the heterogeneity of this study, considering that an emergent exploratory laparotomy is a less severe intervention compared to above knee amputation which is a major surgery that almost always requires intraoperative blood transfusion. Third, the relatively small sample size of new onset POAF cases also limits the strength and generalizability of conclusions. Fourth, we acknowledge that the outcomes and covariates that were used in the models can be influenced by multiple confounders, including baseline comorbidities, illness severity, and complications, and this could potentially affect the findings of this study. Furthermore, while a weak association between IVF intake and POAF was identified, the types of intraoperative IVF were not analyzed, which may affect this association. Lastly, this study focuses on short-term postoperative outcomes and long-term outcomes, such as recurrence of AF or mortality after discharge, were not addressed.

CONCLUSION

In this study, we identified that POAF commonly affects critically ill patients following emergent non-cardiac surgeries (13.0%), presents as new onset POAF in 7.9%, and is associated with poor outcomes. Advanced age, identifying as white, high IVF intake, and extended ICU stay were significant predictors of POAF occurrence. New onset POAF was associated with advanced age and increased risk of cardiac arrest and mortality. Extrapolating from data that show only moderate prediction ability of scores for POAF in cardiac surgery and limited data in non-cardiac surgery, further research is needed to identify patients at highest risk for developing POAF. These findings underscore the need for vigilant monitoring and tailored management strategies in high-risk patients. Future multicenter and prospective studies are warranted to develop targeted interventions aimed at reducing POAF incidence and improving patient outcomes in this population.