Secondary Logo

Predictors, Prognosis, and Management of New Clinically Important Atrial Fibrillation After Noncardiac Surgery

A Prospective Cohort Study

Alonso-Coello, Pablo, MD, PhD*; Cook, Deborah, MD, MSc†‡; Xu, Shou Chun, MD§; Sigamani, Alben, MD; Berwanger, Otavio, MD; Sivakumaran, Soori, MD#; Yang, Homer, MD**; Xavier, Denis, MD, MSc††; Martinez, Luz Ximena, MD‡‡; Ibarra, Pedro, MD§§; Rao-Melacini, Purnima, MSc‖‖; Pogue, Janice, PhD‖‖; Zarnke, Kelly, MD, MSc¶¶; Paniagua, Pilar, MD, PhD##; Ostrander, Jack, MD***; Yusuf, Salim, MBBS, PhD†‡†††; Devereaux, P. J., MD, PhD†‡††† on behalf of the POISE Investigators

doi: 10.1213/ANE.0000000000002111
Patient Safety: Original Clinical Research Report
Free

BACKGROUND: Despite the frequency of new clinically important atrial fibrillation (AF) after noncardiac surgery and its increased association with the risk of stroke at 30 days, there are limited data informing their prediction, association with outcomes, and management.

METHODS: We used the data from the PeriOperative ISchemic Evaluation trial to determine, in patients undergoing noncardiac surgery, the association of new clinically important AF with 30-day outcomes, and to assess management of these patients. We also aimed to derive a clinical prediction rule for new clinically important AF in this population. We defined new clinically important AF as new AF that resulted in symptoms or required treatment. We recorded an electrocardiogram 6 to 12 hours postoperatively and on the 1st, 2nd, and 30th days after surgery.

RESULTS: A total of 211 (2.5% [8351 patients]; 95% confidence interval, 2.2%–2.9%) patients developed new clinically important AF within 30 days of randomization (8140 did not develop new AF). AF was independently associated with an increased length of hospital stay by 6.0 days (95% confidence interval, 3.5–8.5 days) and vascular complications (eg, stroke or congestive heart failure). The usage of an oral anticoagulant at the time of hospital discharge among patients with new AF and a CHADS2 score of 0, 1, 2, 3, and ≥4 was 6.9%, 10.2%, 23.0%, 9.4%, and 33.3%, respectively. Two independent predictors of patients developing new clinically important AF were identified (ie, age and surgery). The prediction rule included the following factors and assigned weights: age ≥85 years (4 points), age 75 to 84 years (3 points), age 65 to 74 years (2 points), intrathoracic surgery (3 points), major vascular surgery (2 points), and intra-abdominal surgery (1 point). The incidence of new AF based on scores of 0 to 1, 2, 3 to 4, and 5 to 6 was 0.5%, 1.0%, 3.1%, and 5.3%, respectively.

CONCLUSIONS: Age and surgery are independent predictors of new clinically important AF in the perioperative setting. A minority of patients developing new clinically important AF with high CHADS2 scores are discharged on an oral anticoagulant. There is a need to develop effective and safe interventions to prevent this outcome and to optimize the management of this event when it occurs.

From the *Biomedical Research Institute Sant Pau (IIB Sant Pau), CIBER Epidemiología y Salud Pública (CIBERESP), Barcelona, Spain; Departments of Medicine and Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada; §Hypertension League Institute, Beijing, China; Department of Clinical Research, Narayana Hrudyalaya Limited, Bangalore, India; Research Institute HCor (Heart Hospital-Hospital do Coracao), Sao Paulo, Brazil; #Department of Medicine, University of Alberta, Edmonton, Alberta, Canada; **Department of Anaesthesia, University of Ottawa, Ontario, Canada; ††St John’s Medical College and St John’s Research Institute, Bangalore, India; ‡‡Department of Medicine, Universidad Autónoma de Bucaramanga, Bucaramanga, Colombia; §§Department of Anaesthesia, Clinica Reina Sofia, Bogota, Colombia; ‖‖Population Health Research Institute, McMaster University, Hamilton, Ontario, Canada; ¶¶Department of Medicine, University of Calgary, Alberta, Canada; ##Department of Anesthesiology, Hospital de la Sta Creu i Sant Pau, Barcelona, Spain; ***Department of Medicine, Grey Bruce Health Sciences, Owen Sound, Ontario, Canada; and †††Population Health Research Institute, McMaster University, Hamilton, Ontario, Canada.

Accepted for publication March 7, 2017.

Funding: Funding for this study was received from the Canadian Institutes of Health Research, the National Health and Medical Research Council of the Commonwealth Government of Australia, the Instituto de Salud Carlos III (Ministerio de Sanidad y Consumo) in Spain, the British Heart Foundation, and AstraZeneca, which provided the study drug and funding for drug labeling, packaging, and shipping, and helped support the cost of some national POISE investigator meetings.

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

This report describes human research. All participating sites obtained ethical approval from institution ethics review boards before recruiting patients.

This report describes cohort observational clinical study. The authors state that the report includes every item in the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for cohort observational clinical studies.

This manuscript was screened for plagiarism using CrossRefMe.

Reprints will not be available from the authors.

Address correspondence to Pablo Alonso-Coello, MD, PhD, Biomedical Research Institute Sant Pau (IIB Sant Pau), CIBER Epidemiología y Salud Pública (CIBERESP), Hospital de la Santa Creu i Sant Pau, Pavellón 18, planta baja, c/ Sant Antoni M. Claret, 167, 08025 Barcelona, Spain. Address e-mail to palonso@santpau.cat.

Worldwide, about 200 million adults undergo major noncardiac surgery annually.1 New atrial fibrillation (AF) is the most common cardiac arrhythmia after noncardiac surgery.2 However, its pathophysiology is not well understood, with several perioperative factors likely involved, including increased sympathetic outflow, metabolic alterations (eg, hypoglycemia/hyperglycemia and electrolyte disturbances), and inflammation.3 A large prospective cohort study (4181 patients) including consecutive patients undergoing noncardiac surgery suggested that 4.1% of patients develop new AF after noncardiac surgery.2 These data suggest that globally, more than 8 million adults may develop new AF after noncardiac surgery annually.

We undertook the PeriOperative ISchemic Evaluation (POISE) trial.4 In this trial, 190 centers in 23 countries randomized 8351 patients with, or at risk for, atherosclerotic disease to receive extended-release metoprolol succinate (metoprolol CR) or placebo starting 2 to 4 hours before surgery and continuing for 30 days. In POISE, 2.5% of patients developed new clinically important AF in the first 30 days, and metoprolol reduced the risk of this outcome (relative risk, 0.76; 95% confidence interval [CI], 0.58–0.99).4 In the primary POISE publication, we reported the results of a model to predict the 30-day risk of stroke after noncardiac surgery. New clinically important AF was an independent predictor of stroke at 30 days (adjusted odds ratio, 3.51; 95% CI, 1.45–8.52).4

Despite the frequency of new clinically important AF after noncardiac surgery and its increased association with the risk of stroke at 30 days, data are limited informing the prediction, associated outcomes, and management of this perioperative complication. Thus, we undertook this POISE substudy, which aimed to derive a clinical prediction rule for new clinically important AF in patients undergoing noncardiac surgery, to determine the associations with outcomes (eg, length of hospital stay, as a proxy for health care resource use), and to assess management (eg, anticoagulation therapy) of these patients.

Back to Top | Article Outline

METHODS

Institutional Review Board and Patient Consent

All participating sites obtained ethical approval from institutional ethics review boards before recruiting patients. All participants or surrogate decision makers provided written informed consent.

Back to Top | Article Outline

Clinical Trial Registration

This trial was registered at ClinicalTrials.gov (registration number: NCT00182039;

https://clinicaltrials.gov/ct2/show/NCT00182039?term=NCT00182039&rank=1).

Back to Top | Article Outline

Design

This was an observational analysis of patients who participated in the POISE trial. Our methodology report and main publication contain additional details.4,5

Back to Top | Article Outline

Patients

Between October 2002 and July 2007, we recruited patients at 190 hospitals in 23 countries. Eligible patients underwent noncardiac surgery, were ≥45 years of age, had an expected length of hospital stay of ≥24 hours, and had or were at risk for atherosclerotic disease.4

Back to Top | Article Outline

Follow-up

Patients were followed throughout their hospitalization and contacted at 30 days after randomization. The 30-day follow-up for clinical outcomes was complete for 8331 (99.8%) of the 8351 randomized patients. We recorded an electrocardiogram (ECG) 6 to 12 hours postoperatively and on the 1st, 2nd, and 30th days after surgery.

Back to Top | Article Outline

Outcomes

The occurrence of new clinically important AF within 30 days of randomization was an a priori–defined secondary outcome that was evaluated in all patients. New clinically important AF was defined as new AF that resulted in angina, congestive heart failure, symptomatic hypotension, or that required treatment with a rate-controlling drug, antiarrhythmic drug, or electrical cardioversion.

Back to Top | Article Outline

Statistical Analyses

Baseline Characteristics.

We determined the proportion of patients in the POISE cohort developing new clinically important AF and the associated 95% CIs. We compared baseline characteristics between the population who experienced new clinically important AF and those who did not using a Student t test for continuous variables and a χ2 test or Fisher exact test for dichotomous variables, as appropriate.

Back to Top | Article Outline

New Clinically Important Atrial Fibrillation Predictors.

We conducted multivariable logistic regression analysis in which the dependent variable was the development of new clinically important AF. Candidate independent variables assessed for possible inclusion in the prediction rule were patients’ baseline characteristics (age [by decile increase], sex, coronary artery disease, peripheral arterial disease, atherothrombotic stroke, hospitalization for congestive heart failure within 3 years of randomization, history of congestive heart failure, diabetes treated with an oral hypoglycemic agent or insulin, preoperative serum creatinine >175 μmol/L [>2.0 mg/dL], history of a transient ischemic attack, emergent/urgent surgery, history of hypertension and smoking status), type of surgery, and type of anesthesia/analgesia, preoperative medications (which included diuretics, amiodarone, calcium channel blockers, digoxin, statins, and angiotensin-converting enzyme/angiotensin-receptor blockers), and treatment allocation to metoprolol or placebo.

Back to Top | Article Outline

Model Development and Validation.

We performed the bootstrap procedure by resampling 1000 samples with replacements and automated backward elimination to determine the predictors of new clinically important AF and to develop a predictive model in our derivation cohort.6 The resampling was done from a derivation data set of 5568 individuals, a random selection of two-thirds of the original data. Backward elimination with the Wald test P ≤ .05 was used as the threshold for keeping a variable in the model. We undertook a sensitivity analysis using forward selection, with the P ≤ .05 for the score χ2 statistic as the threshold for entering a variable in the model. As a proxy for external validity, we divided the 190 centers into two-thirds (126 centers) and one-third (64 centers) and treated them as derivation and validation cohorts, respectively, and we established the scoring in both of these cohorts.

We determined for each candidate variable the proportion of bootstrap samples in which that variable was identified as an independent predictor of the outcome, and we included all variables that were identified as significant predictors of new clinically important AF in at least 60% of the bootstrap samples in our derivation model. We then evaluated the variables from our derivation model in the validation data set (remaining one-third of the original data). For both the derivation and validation models, discrimination was assessed through a c-index and calibration through the Hosmer-Lemeshow (H-L) G statistic.6 Calibration was also assessed graphically by plotting the predicted probability from the model adjusted by the final covariates against the observed probability.

Back to Top | Article Outline

Development of Scoring System.

Finally, we developed a scoring system by assigning weighted points to each statistically significant predictor in our derivation cohort, and the expected 30-day risk of clinically important AF was determined using the method outlined by Sullivan et al.7

Back to Top | Article Outline

Thirty-day Outcomes.

Thirty-day outcomes were compared between the groups of patients who did and did not develop new clinically important AF and were expressed as hazard ratios (or relative risk, when time to event was not available) with 95% CIs. Linear regression analysis was undertaken in which length of stay was the dependent variable, and the perioperative events (ie, congestive heart failure, coronary revascularization, stroke, significant bleeding, clinically important bradycardia, nonfatal cardiac arrest, myocardial infarction, and new clinically important AF) were the independent variables.

Back to Top | Article Outline

Management of New Clinically Important Atrial Fibrillation.

Regarding perioperative medications, we compared the medications administered in the preoperative, in-hospital, and postdischarge periods. Generalized estimating equations for repeated measures were used to assess variations in the usage of cardiovascular drugs across the 3 evaluated time periods (before surgery, during hospital stay, and at hospital discharge).

Additionally, we stratified patients with new clinically important AF according to their CHADS2 score.8 The CHADS2 model predicts the risk of stroke in patients with AF and includes the following characteristics: congestive heart failure, history of hypertension, age ≥75 years, diabetes, and previous cerebral ischemia. The model assigns 1 point for each characteristic except for previous cerebral ischemia, which receives 2 points. By strata, we calculated the proportion of patients who received oral anticoagulants at hospital discharge.

For regression models, we report odds ratios for logistic regression or hazard ratios for Cox proportional hazard regression, as appropriate, with 95% CIs and associated Pvalues. For all tests, we used 2-tailed α = .05 level of significance. We performed all analyses using SAS, version 9.2 for UNIX (SAS Institute, Cary, NC).

Back to Top | Article Outline

RESULTS

A total of 211 (2.5%; 95% CI, 2.2%–2.9%) patients developed new clinically important perioperative AF within 30 days of randomization. Most patients (94.7%) who developed new clinically important AF did so during their index hospitalization.

Multivariable logistic regression in our derivation cohort showed that age (65–74, 75–84, and ≥85 years) and type of surgery (major vascular, intra-abdominal, and intrathoracic) were independent predictors of patients developing new clinically important AF in 60% of the bootstrap samples. The study drug (metoprolol) in the derivation cohort when adjusted by the other covariates had an odds ratio (OR) of 0.75, 95% CI, 0.57–1.00 and was an independent predictor in only 20% of the bootstrap models. Using the forward selection, intra-abdominal surgery was not a significant predictor of new AF in 60% of the bootstrap samples. However, since it was a predictor with the backward selection, it remained in the model for the final calculation of the score. The frequency with which each candidate variable was selected using backward selection in 1000 bootstrap samples drawn from the POISE derivation data set is reported in Appendix 1. The same variables also significantly predicted new clinically important AF when dividing the data set by hospitals in the derivation cohort (see Appendix 2).

Table 1

Table 1

The first model demonstrated a C statistic of 0.69, and the goodness-of-fit test demonstrated a P = .87. Appendix 2 also reports the results of the validation cohort. In this smaller cohort of patients age ≥85 years, intrathoracic surgery and intra-abdominal surgery were not statistically significant independent predictors of clinically important AF. This model had a C statistic of 0.72, and the Pvalue from the H-L goodness-of-fit test was .96. As a measure of external validity, when the derivation cohort consisted of two-thirds of the hospitals, the C statistic was 0.69, and the H-L Pvalue was .82. In the validation cohort with one-third of the hospitals, the C statistic was 0.71, and the Pvalue from H-L test was .37. The calibration plot of the predicted and observed event probabilities in the validation cohort showed reasonable agreement, except for a few under estimation of the event rates (Appendix 3).

Table 2

Table 2

Table 3 reports the derived clinical prediction rule for the outcome of new clinically important AF. Based on the scoring system we developed, patients were classified into 4 risk categories: category 1 (0–1 point), category 2 (2 points), category 3 (3–4 points), and category 4 (5–6 points). In our model, patients with 0 to 1 points had a risk of 0.5% of new clinically important AF, whereas patients with 5 to 6 points had a risk of 5.3% (Table 3).

Table 3

Table 3

Table 4 reports the outcomes of patients who did and did not develop new clinically important AF. Patients developing new clinically important AF developed more vascular complications (nonfatal cardiac arrest, congestive heart failure, myocardial infarction, stroke, coronary revascularization, significant bleeding, clinically important bradycardia, clinically important hypotension, and all-cause mortality). A linear regression model demonstrated that developing new clinically important AF was independently associated with an increase in the length of hospital stay by 6.0 days (95% CI, 3.5–8.5 days).

Table 4

Table 4

Table 5 reports the cardiovascular medication usage preoperatively, during hospital stay, and at hospital discharge among patients who developed new clinically important AF. These patients had a substantial increase in the usage of heart rate–controlling drugs (nonstudy β blockers, calcium channel blockers, and digoxin) and an antiarrhythmic drug (ie, amiodarone) while in hospital, with more than 90% receiving one of these drugs at some point during this time period.

Table 5

Table 5

Among patients who developed clinically important AF, there was an increased usage of antithrombotic drugs (acetylsalicylic acid, clopidogrel/ticlopidine, or oral anticoagulants) during their hospital stay and at discharge compared to the preoperative period. Nevertheless, 38.9% of patients who developed clinically important AF did not receive an antiplatelet or anticoagulant at hospital discharge.

Table 6

Table 6

The usage of an oral anticoagulant at the time of hospital discharge among patients with new clinically important AF and a CHADS2 score of 0, 1, 2, 3, and ≥4 was 6.9%, 10.2%, 23.0%, 9.4%, and 33.3%, respectively (Table 6). Among the 211 patients who developed new clinically important AF, we obtained a 30-day ECG on 120 of them (ie, 57.0%). Twenty (16.6%) of these patients were in AF on their 30-day ECG. Table 6 reports their use of anticoagulants based on their CHADS2 score. A minority of the patients in AF at 30 days after surgery with a high CHADS2 risk score (ie, ≥2) were discharged from hospital on an anticoagulant (ie, 3 of 11 patients; 27.3%).

Back to Top | Article Outline

DISCUSSION

Principal Findings

We developed a clinical prediction model that could prove useful in predicting and targeting patients for prevention of new clinically important AF in patients undergoing noncardiac surgery. New clinically important AF was associated with a poor prognosis and resulted in substantially longer hospital stays. Only a minority of these patients received an oral anticoagulant at hospital discharge, including those with high CHADS2 scores.

Back to Top | Article Outline

Strengths and Limitations of Our Study

Strengths of our study include our evaluation of the largest sample to date of patients undergoing a broad range of noncardiac surgeries in 190 centers in 23 countries.4 We used pretested criteria for comprehensive outcome assessment, rigorous methodology, and we report results using STROBE criteria. We obtained similar results when stratifying data by different methods (two-thirds of the data versus one-third and two-thirds of the centers versus one-third).

Our study also has limitations. POISE included only patients with, or at risk for, atherosclerotic disease. A moderate number of patients developed new clinically important AF, but we do not have long-term outcomes. We did not collect data on anticoagulant usage at the 30-day follow-up, and it is possible that some of the patients who were still in AF at 30 days were started on an oral anticoagulant after hospital discharge. Continual heart rhythm data were not collected to determine the incidence of silent AF, which may be prognostically relevant. We do not know whether patients with resolved AF at 30 days recurred or if they had paroxysmal AF subsequently. Similarly, we do not know the exact time that AF began. Therefore, we cannot differentiate whether new clinically important AF precipitated or resulted from angina, congestive heart failure, or myocardial infarction. Three independent predictors of clinically important AF in our derivation cohort were not statistically significant predictors in our validation cohort. Our prediction model requires validation in an independent cohort

Back to Top | Article Outline

Our Study in Relation to Other Studies

The incidence of new clinically important AF was lower (2.5%) in our study compared to the largest previous cohort study (4.1%) and to the average of the prospective cohorts reviewed by Walsh et al2,9 (4.4%). There are several potential explanations for this difference. Our cohort study was international and included a more representative sample of surgeries compared to prior studies. Also, half of the patients in our study received metoprolol, which reduced the risk of new clinically important AF. Our definition of new clinically important AF was more restrictive than that of earlier studies. Our use of the term new clinically important AF was not meant to imply that all other episodes of perioperative AF are unimportant. We believed that new AF that fulfilled our diagnostic criteria was important, but this would not preclude expansion of our diagnostic criteria based on future data. It is also possible that the actual incidence of AF in this setting might have decreased due to a change of practice (most previous studies were undertaken more than a decade ago).

The independent predictors of new clinically important AF in our cohort are similar to those observed in 2 previous cohorts, and so is the observed increased length of hospital stay associated with perioperative AF.2,10 Other studies have shown similar associations between perioperative AF and outcomes, like death and major cardiovascular outcomes, as we demonstrated in our study.2,10 Two other studies have developed models to predict AF in the noncardiac surgical context; however, these models were developed from administrative databases and were restricted to thoracic surgery patients and lung cancer surgery patients.11,12 The models had moderate accuracy in predicting AF and included different factors (male gender, heart rate ≥72 bpm, African American race, pneumonectomy, bilobectomy, and clinical cancer stage II and above) compared to the predictors identified in our analyses except for age.11,12 These models have not been validated.

Back to Top | Article Outline

Interpretation

Our model predicted the risk of new clinically important AF moderately well in patients undergoing noncardiac surgery; it requires replication in an independent sample. If validated prospectively, our model may prove useful to stratify patients according to their risk and allow physicians to target prevention in high-risk patients. In our sensitivity analysis, intra-abdominal surgery that was based upon a model using forward selection was not an independent predictor of new clinically important AF. Therefore, this risk factor should be viewed with caution. To further enhance risk prediction in the highest risk category, it is possible that laboratory data beyond the clinical history will be required (eg, N-terminal pro-B-type natriuretic peptide).13

In the context of noncardiac surgery, for patients with, or at risk for, atherosclerotic disease, new clinically important AF is a frequent complication. Unfortunately, there is scarce evidence about how to prevent new clinically important AF in patients undergoing noncardiac surgery.14 Colchicine, an anti-inflammatory drug that has shown promise in preventing atrial fibrillation in cardiac surgery is being evaluated in a multicenter pilot.15

This arrhythmia in the surgical context is associated with a poor prognosis and longer hospital stay, which would substantially impact cost. A substantial proportion of patients, even with high CHADS2 scores, who developed new clinically important AF did not receive an oral anticoagulant. This was the practice before we demonstrated that new clinically important AF was an independent predictor of the risk of stroke at 30 days. Further research is needed to evaluate the risk and benefits of antithrombotic therapy in this patient population. Studies, with continuous rhythm monitoring, are also needed after noncardiac surgery to determine the incidence and impact of subclinical AF.

Back to Top | Article Outline

APPENDIX 1.

Frequency With Which Each Candidate Variable Was Selected Using Backward Model Selection in 1000 Bootstrap Samples Drawn From the POISE Derivation Data Set of 5568 Individuals

Table

Table

Back to Top | Article Outline

APPENDIX 2.

Independent Predictors of New Clinically Important Atrial Fibrillation Derivation and Validation Cohorts (Based on Two-thirds and One-third of Randomly Selected Centers) and Points Assigned in the Clinical Prediction Rule

Table

Table

Back to Top | Article Outline

APPENDIX 3.

Calibration Plots of the Predicted Versus the Observed Proportions of New Clinically Important Atrial Fibrillation

Figure

Figure

Calibration plot of the predicted versus the observed new clinically important atrial fibrillation in the derivation cohort, as estimated by the covariates, age, major vascular surgery, and intra-abdominal and intrathoracic surgery. The solid line refers to the exact agreement between observed and predicted event rates. Larger circles indicate greater proportion of events.

Figure

Figure

Calibration plot of the predicted versus the observed new clinically important atrial fibrillation in the validation cohort, as estimated by the covariates, age, major vascular surgery, and intra-abdominal and intrathoracic surgery. The solid line refers to the exact agreement between observed and predicted event rates. Larger circles indicate greater proportion of events.

Back to Top | Article Outline

DISCLOSURES

Name: Pablo Alonso-Coello, MD, PhD.

Contribution: This author helped design the study, conduct the study, and analyze the data, and wrote the manuscript.

Conflicts of Interest: None.

Name: Deborah Cook, MD, MSc.

Contribution: This author helped design the study, conduct the study, and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Shou Chun Xu, MD.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Alben Sigamani, MD.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Otavio Berwanger, MD.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Soori Sivakumaran, MD.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Homer Yang, MD.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Denis Xavier, MD, MSc.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Luz Ximena Martinez, MD.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Pedro Ibarra, MD.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: Pedro Ibarra, MD, received honoraria from Fresenius Kabi for lecture about perioperative fluid administration about $2000.

Name: Purnima Rao-Melacini, MSc.

Contribution: This author helped design the study, analyzed the data, and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Janice Pogue, PhD.

Contribution: This author helped design the study, analyze the data, and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Kelly Zarnke, MD, MSc.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Pilar Paniagua, MD, PhD.

Contribution: This author helped design the study, conducted the study, and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Jack Ostrander, MD.

Contribution: This author helped design the study, conduct the study, and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: Salim Yusuf, MBBS, PhD.

Contribution: This author helped design the study and provided critical revisions to the manuscript.

Conflicts of Interest: None.

Name: P. J. Devereaux, MD, PhD.

Contribution: This author helped design the study, analyzed the data, wrote the manuscript, and provided critical revisions to the manuscript.

Conflicts of Interest: None.

This manuscript was handled by: Richard C. Prielipp, MD.

Back to Top | Article Outline

REFERENCES

1. Weiser TG, Regenbogen SE, Thompson KD, et al. An estimation of the global volume of surgery: a modelling strategy based on available data. Lancet. 2008;372:139–144.
2. Polanczyk CA, Goldman L, Marcantonio ER, Orav EJ, Lee TH. Supraventricular arrhythmia in patients having noncardiac surgery: clinical correlates and effect on length of stay. Ann Intern Med. 1998;129:279–285.
3. Bessissow A, Khan J, Devereaux PJ, Alvarez-Garcia J, Alonso-Coello P. Postoperative atrial fibrillation in non-cardiac and cardiac surgery: an overview. Thromb Haemost. 2015;(suppl 1):S304–S312.
4. Devereaux PJ, Yang H, Yusuf S, et al. Effects of extended-release metoprolol succinate in patients undergoing non-cardiac surgery (POISE trial): a randomised controlled trial. Lancet. 2008;371:1839–1847.
5. Devereaux PJ, Yang H, Guyatt GH, et al. Rationale, design, and organization of the PeriOperativeISchemic Evaluation (POISE) trial: a randomized controlled trial of metoprolol versus placebo in patients undergoing noncardiac surgery. Am Heart J. 2006;152:223–230.
6. Austin PC, Tu JV. Bootstrapping methods for developing predictive models. Am Stat. 2004;58:131–137.
7. Sullivan LM, Massaro JM, D’Agostino RB Sr. Presentation of multivariate data for clinical use: the Framingham Study risk score functions. Stat Med. 2004;23:1631–1660.
8. Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA. 2001;285:2864–2870.
9. Walsh SR, Tang T, Wijewardena C, Yarham SI, Boyle JR, Gaunt ME. Postoperative arrhythmias in general surgical patients. Ann R Coll Surg Engl. 2007;89:91–95.
10. Goldman L. Supraventricular tachyarrhythmias in hospitalized adults after surgery. Clinical correlates in patients over 40 years of age after major noncardiac surgery. Chest. 1978;73:450–454.
11. Onaitis M, D’Amico T, Zhao Y, O’Brien S, Harpole D. Risk factors for atrial fibrillation after lung cancer surgery: analysis of the Society of Thoracic Surgeons general thoracic surgery database. Ann Thorac Surg. 2010;90:368–374.
12. Passman RS, Gingold DS, Amar D, et al. Prediction rule for atrial fibrillation after major noncardiac thoracic surgery. Ann Thorac Surg. 2005;79:1698–1703.
13. Cardinale D, Colombo A, Sandri MT, et al. Increased perioperative N-terminal pro-B-type natriuretic peptide levels predict atrial fibrillation after thoracic surgery for lung cancer. Circulation. 2007;115:1339–1344.
14. Bessissow A, Khan J, Devereaux PJ, Alvarez-Garcia J, Alonso-Coello P. Postoperative atrial fibrillation in non-cardiac and cardiac surgery: an overview. J Thromb Haemost. 2015;13 (suppl 1):S304–S312.
15. Devereaux PJ. Available at: https://clinicaltrials.gov/ct2/show/NCT01985425?term=COP-AF&rank=1. 2015. Accessed August 2, 2016.
© 2017 International Anesthesia Research Society