Demographic and clinical predictors of post-operative atrial fibrillation in cardio-surgical patients

Abstract

Aims

Post-operative atrial fibrillation is defined as an episode of atrial fibrillation that occurs 1–5 days after a surgical procedure in patients without a previous history of atrial fibrillation. Multiple factors such as demographics, cardiac surgical, endogenous, or mental health may relate to post-operative atrial fibrillation.

The aim of this study was to identify multivariable high-risk factors for post-operative atrial fibrillation and to propose a risk-assessment tool.

Methods and results

A cross-sectional observational study was conducted in a University Hospital of Greece. Predictor variables examined demographic and clinical variables, anxiety, depression, health-related quality of life, frailty, perioperative mortality (European System for Cardiac Operative Risk Evaluation II), and 10-year survival/mortality risk (Charlson Comorbidity Index score). The outcome variable was post-operative atrial fibrillation. Multivariable analysis was assessed to identify predictors of post-operative atrial fibrillation.

Ninety-one patients were included in our sample. Post-operative atrial fibrillation was diagnosed in 44 (48.4%). Factors associated with post-operative atrial fibrillation are the following: age group of 66–75 years [OR 5.78, 95% confidence interval (CI) 1.37–24.34], Charlson Comorbidity Index score (OR 1.42, 95% CI 1.07–1.89), and hours of mechanical ventilation (OR 1.03, 95% CI 1.00–1.06). The Charlson Comorbidity Index score was identified as an independent predictor of post-operative atrial fibrillation (exp: 1.412, 95% CI: 1.017–1.961).

Conclusion

Patients with post-operative atrial fibrillation had a higher Charlson Comorbidity Index score. The Charlson Comorbidity Index was identified as an independent clinical predictor of post-operative atrial fibrillation. The risk-assessment tool proposed includes age, Charlson Comorbidity Index score, and hours of mechanical ventilation. Future studies are needed to establish such an assessment.

Graphical Abstract

Open in new tabDownload slide

Introduction

Post-operative atrial fibrillation (POAF) is defined as an episode of atrial fibrillation (AF) that occurs 1–5 days after a cardio-surgical procedure in patients without a previous history of AF. Often conversion to sinus rhythm is spontaneous within 24 h.^1–3 Post-operative atrial fibrillation is associated with increased morbidity and mortality, despite contemporary improvements in cardiac surgery-associated morbidity and mortality.^2,4,5

The incidence of POAF varies from 25 to 50%⁶ and it is significantly higher after cardiac surgeries than in both thoracic surgery (10–30%) and non-cardiac, non-thoracic surgery (1–15%).¹

As the global population grows older, cardiac operations are often performed in older, more frail patients with severe comorbidities such as hypertension, dyslipidaemia, and diabetes, which might be associated with POAF.^7–11

Multiple risk factors have been related to POAF including demographics (i.e. older age, male gender), cardiac surgical factors (i.e. isolated or combined procedures), 25% after coronary artery bypass grafting (CABG), up to 30% after isolated valvular procedures, and 40–50% undergoing both CABG and heart valve surgery,¹ cardio-pulmonary bypass time (CPB), aortic cross-clamp time, and mechanical ventilation time.^7–9 Endogenous (inflammatory pathways, oxidative stress, and autonomic dysfunction) and exogenous post-operative factors might also be related.^9,12

Furthermore, findings in the literature reveal an association between anxiety, depression, health-related quality of life (HRQL), and the incidence and clinical prognosis of AF in the community.^6,13–16 In older, frail patients, who are more vulnerable to the development of AF, anxiety disorders are among the most common mental health (MH) problems.¹⁷

However, most previous studies suggest assessment of risk factors for POAF, from a medical standpoint,^18,19 while only one study has examined the association between MH and POAF.²⁰ It is necessary to develop an holistic, person-centered validated risk-assessment tool, for identifying patients prone to POAF that can be used by all health care professionals.

Το fill the gap, the aim of this study was to identify multivariable high-risk factors for POAF and to propose a risk-assessment tool, including all identified risk factors, for the prevention and early detection of POAF.

Methods

This study was a cross-sectional observational study conducted in the Cardiac Surgery Department of a University Hospital in Greece between December 2018 and March 2020. The study protocol was approved by the human research ethics committee of the University Hospital of Heraklion (No.: 13711/24-09-18), conforming to the principles outlined in the Declaration of Helsinki.²¹

Study population

Eligible participants for this study were all patients over 18 years, scheduled for an elective cardiac operation using cardio-pulmonary bypass, during the aforementioned time period. Patients not speaking Greek language or having a severe mental disorder (schizophrenia, dementia), as well as emergency and re-do cases, were excluded.

Sample size calculations were based on linear regression formula between SF-36 MH and Beck Depression Inventory (BDI) scales. In a previous study, a negative linear relationship (β = −0.39) between SF-36 MH and BDI was found.²² Standard deviations (SDs) for the BDI scale were 2.0, and for SF-36 plus considering α = 0.05 and β = 0.20, a total of 100 patients were considered as the minimum sample size. That sample size was not achieved. However, post hoc estimations showed that the final sample size of 91 patients decreased power by <10%. So, the expected 80% power decreased to 74%.

Data collection, assessments, and variable definitions

Consenting individuals completed an interview with the researcher using a structured questionnaire for socio-demographic characteristics including gender, age, region, occupation, insurance, marital status, and education. The assessment took place 1 day before surgery, after informing them of the study, assuring them of the anonymity of the answers, and giving a written informed consent form to sign. Questionnaires were self-administered and given to the patients and collected the same day by the researcher. The researcher gave instructions to the patients on how to complete the questionnaires in order to avoid inaccuracy or misunderstanding. Furthermore, if any patient needed any help, the researcher provided it. The patients needed ∼45 min to complete the questionnaires.

Additional data such as CPB time (min), aortic cross-clamp time (min), and mechanical ventilation time (h) were retrieved from medical-nursing records and from the local clinical database.

Factors

Post-operative atrial fibrillation was the main outcome variable. Predictor variables of POAF included MH (anxiety symptoms and depression symptoms), HRQL, frailty, perioperative mortality, and survival/mortality risk.

Post-operative atrial fibrillation

Post-operative atrial fibrillation was defined as a new-onset AF occurring 1–5 days after surgery,^1,2,5 detected by the nursing staff, on an electrocardiogram (ECG) monitor/telemetry and confirmed by ECG, with a duration of ≥ 5 min. The POAF was confirmed by the treating physician. Patients’ vital signs were fully monitored during the first post-operative days, and cardiac rhythm was assessed by telemetry for the rest of the hospital stay.

Mental health

Anxiety symptoms: Anxiety symptoms were assessed by the State-Trait Anxiety Inventory (STAI-X-I) scale. STAI-X-I is a self-assessment questionnaire consisting of 40 questions. The first 20, STAI-X-1, include information about how the patient felt while answering the questionnaire (state anxiety, ST) and the remaining 20, STAI-X-2, about how the patient felt overall (trait anxiety, TR).²³ STAI-X-I was developed by Charles Spielberger (1983) and was translated and validated in Greek by Liakos A. and Giannitsi. State trait questions are rated as follows: (i) not at all, (ii) somewhat, (iii) moderately so, (iv) very much so. Trait rate questions are rated as follows: (i) almost never, (ii) sometimes, (iii) often, and (iv) almost always. The mean value for ST is 43.21 and for TR is 42.79, in total 86.01.²⁴

Depression symptoms: Depression symptoms were assessed by the BDI. Beck Depression Inventory is a 21-item self-assessment questionnaire for evaluating the severity of depression in normal and psychiatric populations. Beck Depression Inventory is translated and validated in the Greek language. It demonstrates sufficient validity and satisfactory reliability (Cronbach’s a: 0.84).²⁴

The BDI contains 21 items on a four-point scale from 0 (symptom absent) to 3 (severe symptoms). Scoring is calculated by adding the highest ratings for all 21 items. The minimum score is 0, and the maximum score is 63. In those diagnosed with depression, scores of 0–13 indicate minimal depression, 14–19 (mild depression), 20–28 (moderate depression), and 29–63 (severe depression). The cutoff score, to screen for depression, varied according to the type of sample.²⁵

Beck Depression Inventory and STAI-X-I are copyrighted. The rights are held by Hellenic Institution of Psychology and Health with no fee. They can be used as self-reported assessments or by face-to-face interviews.²⁴ Permission for using both assessments was granted.

Health-related quality of life

Health-related quality of life was assessed by Short Form Health Surveys (SF-36) for physical and MH and quality of life. SF-36 is an indicator of the overall health status. SF-36 has eight scaled scores: physical functioning, role limitations due to physical health problems (RLP), bodily pain (BP), general health perceptions (GH), vitality tapping energy levels and fatigue, social functioning (SF), role limitations due to emotional problems (RLE), and MH.²⁶

The scores are weighted sums of the questions in each section. Scores range from 0 to 100. Lower scores mean more disability, and higher scores mean less disability. As SF-36 is a generic measure, the recommendation would be to focus on the physical and mental scores rather than the overall summary score.^26,27

SF-36 is translated and validated in the Greek language. There is strong evidence that each scale measures a distinct concept. SF-36 can be used as self-reported assessments or by face-to-face interview, and it is free to use.

Frailty

Frailty was assessed by the frail non-disabled (FiND) tool. The FiND questionnaire consists of five questions: two (A and B) are specifically aimed at identifying individuals with mobility disability, and three assess components of the frailty syndrome: weight loss (C), exhaustion (D), and sedentary behaviour (E).²⁸ If A + B ≥ 1, the individual is considered as ‘disabled’, A + B = 0 and C + D + E ≥ 1, the individual is considered as ‘frail’, and A + B + C + D + E = 0, the individual is considered as ‘robust’. The FiND questionnaire presented 95% specificity [95% confidence interval (CI) 75.1–99.2%] and 76% (95% CI 54.9–90.6%) in identifying non-disabled frail participants.^26,29 It can be used as self-reported assessments or by face-to-face interviews.^30,31

Perioperative mortality

Perioperative mortality was assessed by European System for Cardiac Operative Risk Evaluation II (EuroSCORE II). Variables used for the EuroSCORE II calculation are age, gender, renal impairment, pulmonary hypertension, extracardiac arteriopathy, mobility status, previous cardiac surgery, chronic lung disease, active endocarditis, pre-operative state, diabetes mellitus status, New York Heart Association classification, angina at rest, left ventricular systolic function, recent myocardial infarction, urgency for the operation, and weight of the intervention. The median EuroSCORE II value is 1.3%, which indicates low in-hospital mortality. The area under the ROC curve is 0.85 (95% CI 0.75–0.94), suggesting a very good correct classification of the patients. Items are scored on the website in accordance with the logistic EuroSCORE II method to generate a percentage risk estimate.³² It is validated in the Greek population and is free for use on line.³³

Survival/mortality risk

Pre-operatively 10-year survival/mortality risk, in patients with several comorbidities, was assessed by the Charlson Comorbidity Index (CCI). Variables used for the CCI calculation are myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease—rheumatic disease, peptic ulcer disease, liver disease, diabetes, paraplegia and hemiplegia, renal disease, cancer, and HIV/AIDS. Items are scored on the website in accordance with the logistic CCI method to generate a percentage risk estimate.^34,35

Data analysis

Demographic, clinical, and scale-assessed scores were selected and expressed as means and SDs when the variables were continuous, usually questionnaire scores, and as counts, percentages, and proportions when data were in categorical form (quantitative and discrete continuous variables).

Pearson’s χ² was applied to estimate the association between POAF and discrete data.

Differences between groups with and without POAF were assessed in terms of MH (anxiety symptoms and depression symptoms), HRQL, frailty, perioperative mortality, and survival/mortality risk through univariate analysis. Use of crude ORs with 95% CI was also present to establish the risk or protective factors of measured variables to POAF.

After crude analysis, adjusted ORs with 95% CIs were applied using multiple logistic regression. Post-operative atrial fibrillation was the dependent variable (POAF present or not), and the explanatory set included all variables that have a P-value < 0.200 (P < 0.200) in crude analysis of risk/protective effect to POAF.

Missing data

Missing data in any of the questionnaire (SF-36, BDI, SAST, BIS) items do not exceed 5% (0–4 cases). IBM SPSS Statistics 26.0 was used for statistical analysis, and the level of acceptance was set to a = 0.05.

Results

Socio-demographic and clinical characteristics

The final sample size included 91 patients, since from a total of 95 patients, who were planning to undergo an elective cardiac surgical procedure, two refused to participate in the study and two were unable to complete the questionnaires due to older age. They had hearing and visual problems.

The estimated POAF incidence in our study was 44/91 cases = 48.4%. The age group of 66–75 years old presented the highest incidence of POAF (63.4%).

Entire group (total sample)

Most of the participants were males (n = 61, 67.0%), 66–75 years old (n = 41, 45.1%), and residing in urban areas (n = 53, 58.2%). Among comorbidities, hyperlipidaemia (n = 76, 83.5%), hypertension (n = 74, 81.3%), and diabetes mellitus (n = 44, 46.6%) were the most prevalent. In terms of cardiac surgery performed, CABG was performed in 51.6% (n = 47) of the sample, aortic valve replacement (AVR) in 26.4% (n = 24), combined CABG + AVR in 15.4% (n = 14), and mitral valve replacement (MVR) in 6.6% (n = 6) (Table 1).

Group with post-operative atrial fibrillation

Proportions of measured characteristics in POAF/non-POAF patients were not significantly different for comorbidities (hypertension P = 0.338, hyperlipidaemia P = 0.323, and diabetes mellitus P = 0.761) and for the type of surgery (CABG P = 0.593, AVR P = 0.805, CABG + AVR P = 0.317, MVR P = 0421). The distribution of POAF into age groups was 23.1, 38.1, 63.4, and 43.8%, significantly different from non-POAF patients (P = 0.045), although age means and medians did not differ (non-POAF 65.7 ± 11.1, median 66; POAF 69.2 ± 8.7, median 71; P = 0.096) (Table 1).

Comparisons between patients with and without post-operative atrial fibrillation

Univariate analysis of socio-demographics and clinical variables

Patients with POAF compared with patients without POAF were more likely to be in the age group from 66 to 75 years old and had OR 5.78 (CI 1.37–24.34, P = 0.017), which is statistically significant. In terms of comorbidities and type of surgery, comparisons did not reveal any significant differences between the two groups (Table 1).

Furthermore, patients with POAF required a longer period of mechanical ventilation when compared with those without POAF (OR 1.03, CI 1.00–1.06, P = 0.047). Cardio-pulmonary bypass time and aortic cross-clamp time were not different between the groups (Table 2).

Scales assessed

Entire group (total sample)

Mean values of ST and TR for anxiety were 53.1 ± 6.8 and 45.9 ± 6.8, respectively, while BDI for depression showed a mean of 10.3 ± 6.7. The mean FiND score for frailty was 1.4 ± 1.2 and SF-36 scales for HRQL ranged from 41.4 ± 37.1 for RLP to 70.1 ± 31.2 for BP. European System for Cardiac Operative Risk Evaluation II for perioperative mortality showed a mean value of 2.9 ± 2.3 and CCI for comorbidity showed 4.2 ± 1.7, with the estimated %10 years survival of 48.8 ± 30.3% (Table 3).

Group with post-operative atrial fibrillation

All scales completed pre-operatively and their differences in mean scores between patients without and with POAF are presented in Table 3. Mean values of ST and TR for anxiety were 52.6 ± 8.2 and 45.9 ± 7.6, respectively, while BDI for depression showed a mean of 11.4 ± 6.7. The mean FiND score for frailty was 1.6 ± 1.2, and SF-36 scales for HRQL ranged from 38.6 ± 39.8 for RLP to 70.3 ± 34.0 for BP. European System for Cardiac Operative Risk Evaluation II for perioperative mortality showed a mean value of 3.2 ± 2.5 and CCI for comorbidity showed 4.6 ± 1.8, with the estimated %10 years survival of 42.2 ± 30.7% (Table 3).

Comparisons of scales assessed between patients with and without post-operative atrial fibrillation

Univariate analysis of scales assessed

Statistically significant differences were observed in the CCI score [3.7 ± 1.4 vs. 4.6 ± 1.8; P = 0.012; OR 1.42 (CI: 1.07–1.89)] and survival rates [55.1 ± 29.0 vs. 42.2 ± 30.7; P = 0.042; OR 1.01 (CI 1.00–1.03)] (Table 3).

Multivariable analysis, predictors of post-operative atrial fibrillation

In multivariable analyses, the following independent variables were entered together: BDI, EF from SF-36, CCI score, and frailty score. As shown in Table 4, a significant independent predictor of POAF was CCI score (exp: 1.412, CI 1.017–1.961, P = 0.0039). None of the rest of the independent variables entered was significantly associated with POAF.

Discussion

In this study, the incidence of POAF was relatively high. Recent reviews report similar findings.^2–4 Post-operative atrial fibrillation remains a very common complication probably because the existing risk-assessment tools are not sufficient.

Also, patients with POAF had higher CCI scores, lower 10-year survival expectancy, prolonged need for mechanical ventilation, and were more likely to be in the age group of 66–75 years. The CCI score was identified as an independent clinical predictor of POAF.

The first significant finding was that patients with POAF had a higher CCI score and a lower 10-year survival expectancy; thus, CCI was identified as an independent clinical predictor of POAF. One explanation might be that the CCI scoring system is associated with age, comorbidities (i.e. peripheral vascular diseases, cerebrovascular diseases, dementia, chronic obstructive pulmonary disease, diabetes mellitus, chronic kidney diseases), and risk factors, like previous myocardial infarction, which are known risk factors for POAF.^7,8,11 Most studies referring to cardiac surgeries take into consideration comorbidities as entities,^7–11 and less as a score, which make the comparison of findings difficult. Maaroos et al.¹¹ had used the CCI score in their study and reported associations in accordance with this study. They concluded that according to CCI, POAF patients had an almost 10% lower 10-year survival expectancy, as compared with non-POAF patients, prior to surgery. Calculating the CCI score prior to surgery might help to identify patients at risk of developing POAF and plan appropriate interventions to prevent or early detect it [i.e. haemodynamic monitoring, correct electrolyte imbalance (Mg²⁺), pharmacological prophylaxis, maintain fluid balance].^7–11

The second significant finding was that patients with POAF had prolonged mechanical ventilation needs, resulting in delayed extubation. One explanation might be that mechanical ventilation increases atrial pressure and post-operative sympathetic activity, concluding that prolonged ventilation may be an additive mechanism of POAF via these two pathways.

Erdil et al.⁷ had similar findings in their study, reporting that immediate extubation reduces POAF, independently of comorbidities, identifying late extubation as an independent risk factor for POAF. However, Onk et al.⁸ and Karabacak et al.⁹ found no relation.

The new recommendations are to extubate patients by the third hour after admission to the ICU.³⁶ Researchers from 1990 until today embrace this method for its association with decreasing adverse outcomes of prolonged mechanical ventilation. It would be interesting to evaluate the implementation of a nurse-led protocol for early extubation after cardiac surgery in ICU. Furthermore, such a protocol may gradually lead to higher rates of early extubation and thus prevent POAF by decreasing atrial pressure and sympathetic activity.^36–38

In terms of socio-demographic characteristics, patients between 66 and 75 years were more likely to have POAF. Woldendorp et al.,¹⁸ Gudbjartsson et al.,⁴ and Coletta et al.⁵ agreed to this finding, with the third group reporting that age-related changes in the heart, such as increased collagen and loss of ventricular compliance, are responsible for POAF. Sigurdsson et al.⁴⁰ suggested that patients >85 years were a more vulnerable group. The early identification of the age group at higher risk may allow health providers to take pre-operatively, targeted prophylactic measurements and have targeted perioperative management.

Regarding clinical characteristics, no association was found between POAF and type of disease, type of surgery, and comorbidities. Concerning comorbidities, findings are in accordance with Eikelboom et al., Erdil et al., Maaroos et al., and Karabacak et al. Hypertension, dyslipidaemia, and diabetes mellitus were the most common in cardiac patients, and they were observed equally in patients with and without POAF.^7,9–11 Sigurdsson et al., Coletta et al., and Dobrev et al. had different findings reporting that POAF is associated with the aforementioned comorbidities.^40,2,5, Sigurdsson et al. pointed out hypertension as a risk factor for POAF. ⁴⁰They all concluded that the underlying mechanisms of POAF involve a combination of pre-existing factors (i.e. chronic hypertension, diabetes), which often remodel the atria to make it vulnerable to POAF initiation and maintenance along with intraoperative and post-operative changes.

In the present study, no significant difference was observed between the groups with or without POAF in relation to CPB and aortic cross-clamp time, which is consistent with two similar studies.^7,9 An association between POAF and prolonged CPB time had been reported by Sigurdsson et al. and Sandler et al. probably due to post-operative inflammation caused by CPB.^40,34, Onk et al.⁸ identified aortic cross-clamp time as a predictor of POAF.

Finally, no proof was found that MH status (depressive and anxiety symptoms), HRQOL, and frailty were related to POAF. The lack of such associations, in our study, may be due to the elective procedures included as well as the small sample size. Similar findings were reported by Tully et al.²⁰, but there is a lack of studies associating mental status and POAF. On the other hand, there are many studies exploring the association between AF and MH, identifying anxiety as a risk factor.^15,13,39

Importantly, in this study, EuroSCORE II values were not related to POAF. Previous studies have controversial findings. Erdil et al.⁷ concluded that EuroSCORE II can be used as a predictor of POAF because it takes into account general patient information, pre-operative and cardiac risk factors, and operative information. Lack of significance in our study may be due to inadequate sample size to reveal such differences. Furthermore, in elective procedures included in this study, EuroSCORE II might be lower than in emergencies. Future studies, on larger sample sizes, are needed to address these issues.

Limitations

There are some limitations to this study. The main limitation is that the sample size is relatively small, and this may result in a lack of significant correlation to POAF with some of the variables examined (i.e. gender, type of surgery). Initially, a total of 100 patients were considered as the minimum sample size. That sample size was not achieved due to logistical problems within the research team. However, the power of the study did not decrease notably. Future studies with larger samples are required to replicate our findings. Finally, the cross-sectional nature of our study does not allow us to examine causality between related factors.

Conclusions—areas for further studies

Unfortunately, multiple advances in surgery and perioperative care have not led to a reduction in the incidence of POAF in daily clinical practice, which remains high (48.4%). Furthermore, the population undergoing cardiac surgeries is now older, with more comorbidities and relatively higher CCI scores.

According to the findings of the present study, patients in the age group of 66–75 years, with comorbidities expressed as CCI scores and prolonged hours of mechanical ventilation, leading to late extubation, are in high risk of developing POAF.

Recommendations for research include studies with a larger sample size, including not only elective but also emergency surgeries, with a single definition of POAF and more variables in order to establish a risk-assessment tool for POAF. Clinical trials are needed to be conducted in order to evaluate new risk-assessment tools.

Acknowledgements

There were no other contributors to the article than the authors since there was no writing assistance regarding our paper. The authors want to thank all participants for their contribution to this research.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

The data underlying this article are available in the article. Additional detailed data can be requested from the authors.

References

Author notes