CHADS2 Score Predicts Postoperative Atrial Fibrillation in Patients Undergoing Elective Pulmonary Lobectomy
GENERAL THORACIC
Svetlana Kotova, MD, Mansen Wang, PhD, Katie Lothrop, BA, Gary Grunkemeier, PhD, Heather E. Merry, MD, and John R. Handy, MD Division of Thoracic Surgery, Providence Health & Services, Portland; and Medical Data Research Center, Providence Health and Services, Portland, Oregon
Background. Postoperative atrial fibrillation (PAF) affects 12% to 17% of patients undergoing lobectomy and is associated with increased morbidity. CHADS2 (congestive heart failure history, hypertension history, age ‡75 years, diabetes mellitus history, and stroke or transient ischemic attack symptoms previously) is used to predict stroke risk in patients with existing AF. It also has been shown also to predict new-onset PAF. Our objective was to determine whether CHADS2 can predict PAF in patients undergoing lobectomy. Methods. A prospective thoracic surgery clinical database was reviewed to identify adult patients, without prior AF, who underwent elective lobectomy between January 1, 2005, and June 30, 2014. Nonelective and combined operations were excluded. Two groups (PAF and no PAF) were analyzed. Results. PAF developed in 113 of 933 patients with overall incidence of 12% for the entire group. Age (‡75 years) and coronary artery disease were the only significant preoperative characteristics between the two groups. Intensive care unit readmission, new neurologic events, length of stay, 30-day survival, and hospital mortality were
significantly higher in the PAF group as were mean CHADS2 scores (1.4 and 1.1 respectively, p [ 0.0014). Incidence of PAF ranged from 7.9% in low-risk groups to 11% in moderate-risk and 17.7% in high-risk groups, which was also significant, p < 0.0002. Similar findings were noted for CHA2DS2-VASc (age in years, sex, history of congestive heart failure, history of hypertension, history of stroke/transient ischemic symptoms/thromboembolic events, history of vascular disease, history of diabetes mellitus). Conclusions. Although multiple risk factors for PAF have been described, no easily applicable clinical model exists. Observed rate of PAF was significantly lower then the previously described 12% when CHADS2 was 0. CHADS2 can predict PAF in patients undergoing elective lobectomy and can identify patients to selectively institute prophylactic measures in patients at the greatest risk, such as patients with score of 2 or greater. Further validation of this model is warranted in a larger group.
ostoperative atrial fibrillation (PAF) affects 12% to 17% of patients undergoing pulmonary lobectomy [1–5]. PAF has been found to prolong length of hospitalization and intensive care unit use [2, 3, 5–7] and adversely affect short- and long-term mortality [3, 8]. Multiple studies focused on identifying risk factors with the goal of controlling PAF and improving patient outcomes [2–5]. However, despite several univariate and multivariate analyses performed on large patient cohorts, no readily available predictive model exists for individual patient assessment. Although prophylactic measures intuitively make sense, current guidelines do not have
strong recommendations. The Society of Thoracic Surgeons Practice Guidelines state that prophylactic b-blockers are reasonable to use, and this recommendation is based on level 2 evidence [9]; European Association for Cardio-Thoracic Surgery states that there is insufficient evidence to recommend routine prophylactic measures in patients undergoing thoracic operations [10]. Furthermore, none of the published factors predictive of PAF have been incorporated into the guidelines or applied to clinical practice. CHADS2 (history of congestive heart failure, history of hypertension, age >75 years, history of diabetes mellitus, history of stroke or transient ischemic symptoms) is a validated tool used in patients with existing AF to predict risk of stroke and to guide clinical decision regarding initiation and type of anticoagulation [11]. It consists of multiple components each contributing a certain point value to the total score. The following components are taken into account: hypertension, diabetes mellitus, age 75 years or greater, congestive heart failure (all 1 point
P
Accepted for publication Nov 7, 2016. Presented at the Poster Session of the Fifty-second Annual Meeting of The Society of Thoracic Surgeons, Phoenix, AZ, Jan 23–27, 2016. Address correspondence to Dr Kotova, Providence Health & Services, Division of Thoracic Surgery, Ste 6N50, 4805 NE Glisan St, Portland OR 97213; email:
[email protected].
Ó 2017 by The Society of Thoracic Surgeons Published by Elsevier Inc.
(Ann Thorac Surg 2017;103:1566–72) Ó 2017 by The Society of Thoracic Surgeons
0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2016.11.007
Ann Thorac Surg 2017;103:1566–72
Patients and Methods Predetermined patient data points are collected from the two medical centers. This prospective thoracic surgery clinical database was queried to identify all adult patients who underwent elective pulmonary lobectomy between January 1, 2005, and June 3-, 2014. Patients were excluded for age younger than 18 years, nonelective operation, concurrent operations (cardiac, esophageal, spine, abdominal) with pulmonary lobectomy, pulmonary resection other than lobectomy, and preexisting AF. PAF was defined as new-onset AF or flutter detected on telemetry or by electrocardiogram requiring treatment. Patients who experienced PAF comprised a study group, and patients who did not experience PAF comprised a comparison cohort. Baseline characteristics and preoperative and perioperative variables were compared between the two groups using Student t test for continuous variables or c2 (or Fisher exact) test for categorical variables. CHADS2 scores were calculated for each patient, and mean values in each group were compared. Subsequently, entire patient cohort was allocated to low (CHADS2 score 0), moderate (score 1), or high (score 2 or greater) risk groups based on their calculated CHADS2 score. Incidence of PAF was calculated in each of these groups to study CHADS2 score effect and to compare it between the three groups to better understand estimated risk. Logistic regression using multiple variable was defined from the literature. A p value less than 0.05 was considered statistically significant. Analyses were performed in Excel (Microsoft, Redman, WA) for Mac version 15-21-1 and in SAS 9.3 (SAS Institute, Cary, NC) as appropriate. This research project was approved by the Providence Health and Services Institutional Review Board.
Results Patient Population During the stated time interval, a total of 2,157 pulmonary operations were performed with 1,036 being pulmonary
1567
lobectomies, which accounts for 48% of all lung resections. Pneumonectomy, bilobectomy, and segmentectomy constituted a total of 250 operations during the same time frame and were excluded. After exclusion criteria were applied, we identified 933 elective pulmonary lobectomies. Of those, 113 patients experienced PAF (12%) and were compared with 820 patients who did not experience this complication (no PAF group). Post hoc analysis showed the power of the present study to be 0.80 to detect a meaningful difference between the groups.
Preoperative Characteristics Table 1 lists patients demographic, preoperative, and perioperative characteristics. In summary, there was a significant difference related to patients’ age (with mean age in the control group being 7 years younger than the PAF group, p ¼ 0.0001). In addition, patients 75 years and older accounted for a much higher proportion in the PAF group. Proportion of patients with coronary artery disease (CAD) was also higher in the PAF group (23.9% versus 15.9%, p ¼ 0.0322). Forced expiratory volume in 1 second (FEV1) was not different between the groups, but diffusion capacity of the lung for the carbon monoxide (DLCO) was lower in the PAF group. (FEV1 was not available for 7% of patients in the control group and for 3% in the PAF group. DLCO values were missing for a larger number of patients, 17% in the control and 14% in the PAF groups.) Remaining characteristics were all comparable between the two groups. During the study period, proportion of open procedures decreased from 70% to 38% in the no PAF group and from 71% to 43% in the PAF group when the period was divided into 2005 to 2010 and 2011 to 2014 periods, respectively, to evaluate temporal change in the surgical approach. Length of stay was significantly longer in the PAF group (10.4 days versus 6.2 days, p < 0.0001). Furthermore, 30-day survival was 90.3% in PAF patients and was significantly higher at 97.1% in patients who did not experience PAF, (p ¼ 0.0003). In addition, 99% in the control group survived until discharge, whereas 95% were alive at the time of discharge in the PAF group (p ¼ 0.0008).
Postoperative Characteristics We started collecting intensive care unit (ICU) admission and readmission data in 2012, and it was available for 228 patients in the no PAF group and 26 patients in the PAF group. It has been widely practiced to admit patients after lobectomy to the ICU for hemodynamic and respiratory status observation. This is reflected in our study with more than 80% of patients spending the first day in the ICU. This initial admission was 1.5 days in the no PAF group and 1.4 days in the PAF group, which is not significantly different. However, 27% of those patients who experienced PAF during the postoperative period were readmitted to the ICU, compared with only 3% of control patients. Once readmitted, there was no significant difference in additional ICU days between the two groups. We were not able to determine the exact cause of ICU readmission because the exact sequence of events
GENERAL THORACIC
each), and prior history of cerebrovascular events (2 points). Total score corresponds to risk of stroke rate with ability to stratify patients into low-, moderate-, and highrisk categories [12] Although CHADS2 calculation is most widely used to predict embolic complications of existing AF, it has been applied to surgical patients and found to correlate with risk of developing new PAF in patients undergoing cardiac operations [13]. Pulmonary lobectomy is the most common operation performed for lung cancer [1–4, 14]. Patients are usually older with multiple medical comorbidities [1–4], most of which are the same as characteristics taken into CHADS2 calculations. Thus, we aimed to review our database to determine whether CHADS2 score can be a convenient clinical calculator to predict PAF in patients undergoing elective lobectomy. We focused on elective operations because the elective setting gives a clinician an opportunity to institute an intervention.
KOTOVA ET AL POSTOPERATIVE ATRIAL FIBRILLATION
1568
KOTOVA ET AL POSTOPERATIVE ATRIAL FIBRILLATION
Ann Thorac Surg 2017;103:1566–72
Table 1. Baseline, Perioperative, and Postoperative Characteristics of Patients
GENERAL THORACIC
Characteristics Sex, female Age, years Age 75 years Coronary artery disease Cerebrovascular disease Congestive heart failure Diabetes mellitus Hypertension Peripheral vascular disease FEV1, % DLCO, % Smoking history Current Never Prior Steroid use Pathologic process NSCLC Carcinoid Benign Small cell carcinoma Metastatic/other malignancy Not available Perioperative factors Open approach Hospital LOS, days Initial ICU admissiona Initial ICU LOS,a days ICU readmissiona Additional ICU daysa Postoperative complications Air leak >5 days Empyema Bronchopleural fistula Atelectasis Ventilatory support >48 hours Pneumonia Pneumothorax Reintubation ARDS Other pulmonary complications Myocardial infarction Neurologic events Deep venous thrombosis Urinary tract infection Sepsis Blood transfusion Pulmonary embolism 30-Day mortality Mortality at discharge a
No PAF (n ¼ 820)
PAF (n ¼ 113)
447 (54.5) 65.4 10.9 172 (21) 130 (15.9) 60 (7.3) 24 (2.9) 129 (15.7) 461 (56.2) 82 (10) 84 20.3 75 19
60 (53.1) 72 8 46 (40.7) 27 (23.9) 9 (8) 5 (4.4) 19 (16.8) 73 (64.6) 16 (14.2) 83 20.6 70 17.6
193 168 459 28
(23.5) (20.5) (56) (3.4)
25 19 69 6
(22.1) (16.8) (61.1) (5.3)
0.314
685 51 27 9 42 6
(83.5) (6.2) (3.3) (1.1) (5.1) (0.7)
99 5 4 1 4 0
(87.6) (4.4) (3.5) (0.9) (3.6) (0)
0.1886 0.4520 0.9497 0.8372 0.2573 0.7107
475 (57.9) 6.2 4.6 187 (82.0) 1.5 1.45 7 (3.0) 4.1 4.3 145 3 4 48 2 38 14 5 9 29 5 4 5 19 3 17 5 24 9
(17.7) (0.4) (0.5) (5.9) (0.2%) (4.6) (1.7) (0.6) (1.1) (3.5) (0.6) (0.5) (0.6) (2.3) (0.4) (2.1) (0.6) (2.9) (1.1)
p Value 0.7774 0.0001 0.0001 0.0322 0.805 0.3902 0.7681 0.0915 0.1768 0.6263 0.0278 0.7397
70 (61.9) 10.4 11.6 23 (88.5) 1.43 0.84 7 (27.0) 5.4 6.1
0.4169
