2017 Volume 81 Issue 12 Pages 1792-1797
Background: The aim of this study was to assess the accuracy of CHA2DS2-VASc score in predicting new-onset atrial fibrillation (AF) in patients with chronic obstructive pulmonary disease (COPD).
Methods and Results: A total of 50,430 COPD patients were enrolled in this study. The area under the receiver operating characteristic curve (AUC) and the Cox model c-statistic were used to assess the association between new-onset AF risk and CHA2DS2-VASc score in COPD patients. After adjustment for comorbidities other than the components of CHA2DS2-VASc score, risk of new-onset AF in COPD patients increased from 1.24 (95% confidence interval (CI): 1.01–1.52) for a score of 1, to 2.15 (95% CI: 1.62–2.86) for score ≥6 (trend test, P<0.001), compared with CHA2DS2-VASc score 0. The AUC for CHA2DS2-VASc score in predicting new-onset AF in COPD patients was 0.69 (95% CI: 0.68–0.70). The c-statistic of Cox model in predicting incident AF was 0.73.
Conclusions: Risk of new-onset AF in COPD patients increased with increasing CHA2DS2-VASc score. The predictive ability of the score was moderate. CHA2DS2-VASc score might be used as a screening tool for AF in COPD patients.
Chronic obstructive pulmonary disease (COPD), one of the leading causes of death in the world, is associated with cardiovascular disease (CVD).1–4 Atrial fibrillation (AF) is a common arrhythmia and is associated with several cardiovascular morbidities.5–8 The connection between COPD and AF has been proposed; indeed, the negative prognostic impact of AF on patients with COPD has also been reported.9–14
Editorial p 1766
CHA2DS2-VASc score, the clustering of several CVD-related risk factors, was initially proposed for stroke prediction in non-valvular AF.15–17 Recently, the role of CHA2DS2-VASc score has been expanded to predict different adverse cardiac and cerebrovascular outcomes in different populations.18–21 In order to reduce AF-related adverse cardiovascular events in COPD patients, it is essential to construct an accurate, convenient and easy-to-use risk stratification model for prediction of new-onset AF in COPD patients. Using a nationwide cohort, this study assessed the accuracy of CHA2DS2-VASc score in predicting incident AF in this population.
Since 1995, Taiwan’s National Health Insurance (NHI) Bureau has provided a single-payer health insurance program with a coverage rate of >99% of residents in Taiwan.22 The Longitudinal Health Insurance Database 2000 (LHID 2000) of the NHI program was obtained. Detailed descriptions of the LHID 2000 have been reported previously.23,24 The diagnosis of the disease was coded based on the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). This study was approved by the Institutional Review Board (IRB) of China Medical University and Hospital (CMUH104-REC2-115).
Sampled ParticipantsThe subject selection procedure is shown in Figure 1. Between 1 January 2000 and 31 December 2011, the date of diagnosis of COPD (ICD-9-CM codes 491, 492, 496) was defined as the index date. Patients aged ≥18 years at the index date were included in the present study. We excluded patients who had a diagnosis of AF or atrial flutter (ICD-9-CM code 427.31, 427.32) before the index date. CHA2DS2-VASc score was used to estimate the risk of incident AF in patients with COPD.25,26 Comorbidities associated with AF development, such as cancer (ICD-9-CM codes 140–208), hyperlipidemia (ICD-9-CM code 272), chronic kidney disease (ICD-9-CM code 585), hyperthyroidism (ICD-9-CM code 242) and valvular heart disease (ICD-9-CM code 424) were noted. All subjects were followed up until occurrence of new-onset AF, censoring because of death, withdrawal from the NHI program, or at the end of 2011 (whichever occurred first).
Subject selection. AF, atrial fibrillation; COPD, chronic obstructive pulmonary disease.
Demographic data are presented as mean±SD for continuous variables and as number and percentage for categorical variables. The categorical variables and the continuous variables were compared using chi-squared test and Student’s t-test, respectively. The incidence density rate (1,000 person-years) of new-onset AF was estimated according to CHA2DS2-VASc score. Univariable and multivariable Cox proportional hazard regression models were used to assess the risk of new-onset AF in patients with COPD according to CHA2DS2-VASc score. The Cox models were also used to examine the effect of factors of CHA2DS2-VASc score on the risk of incident AF in COPD patients. CHA2DS2-VASc score-specific survival probability for AF was analyzed using Kaplan-Meier survival for actual incidence and Cox modeling for estimated incidence, respectively. Area under the receiver operating characteristic (ROC) curve (AUC) was used to quantify the discriminatory performance of CHA2DS2-VASc score in predicting new-onset AF in COPD patients. Moreover, the c-statistic was also estimated based upon Cox regression analysis.27 All statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC, USA). The significance level for all analyses was set at P<0.05.
Eligible study participants consisted of 50,430 COPD patients; among them, 2,450 patients developed new-onset AF whereas 47,980 patients did not (Table 1). Mean age was 71.6±10.7 years for the AF group and 59.1±16.3 years for the non-AF group. Hypertension was the most common comorbid illness (73.3% vs. 47.5%) in both groups. Mean CHA2DS2-VASc score was 2.85±1.55 for the AF group and 1.84±1.56 for the non-AF group. Mean follow-up period was 4.08±3.00 years for the AF group and 6.91±3.51 years for the non-AF group.
| Variable | AF | P-value | |
|---|---|---|---|
| No (n=47,980) | Yes (n=2,450) | ||
| Age (years) | <0.001† | ||
| ≤64 | 28,378 (59.2) | 581 (23.7) | |
| 65–74 | 11,218 (23.4) | 891 (36.4) | |
| ≥75 | 8,384 (17.5) | 978 (39.9) | |
| Mean±SD | 59.1±16.3 | 71.6±10.7 | <0.001‡ |
| Sex | <0.001† | ||
| Male | 27,298 (56.9) | 1,489 (60.8) | |
| Female | 20,682 (43.1) | 861 (39.2) | |
| Underlying disease (components of CHA2DS2-VASc score) | |||
| CHF | 2,238 (4.66) | 406 (16.6) | <0.001† |
| Diabetes mellitus | 5,825 (12.1) | 328 (13.4) | 0.07† |
| CVA or TIA | 3,127 (6.52) | 235 (9.59) | <0.001† |
| Vascular disease | 2,347 (4.89) | 175 (7.14) | <0.001† |
| Hypertension | 22,792 (47.5) | 1,795 (73.3) | <0.001† |
| Other underlying disease | |||
| Cancer | 1,676 (3.49) | 83 (3.39) | 0.78† |
| Hyperlipidemia | 12,203 (25.4) | 665 (27.1) | 0.06† |
| CKD | 5,411 (11.3) | 413 (16.9) | <0.001† |
| Hyperthyroidism | 1,449 (3.02) | 62 (2.53) | 0.17† |
| Valvular heart disease | 1,171 (2.44) | 131 (5.35) | <0.001† |
| CHA2DS2-VASc score | <0.001† | ||
| 0 | 9,636 (20.1) | 124 (5.06) | |
| 1 | 1,4661 (30.6) | 354 (14.5) | |
| 2 | 9,731 (20.3) | 583 (23.8) | |
| 3 | 6,781 (14.1) | 625 (25.5) | |
| 4 | 3,998 (8.33) | 419 (17.1) | |
| 5 | 1,866 (3.89) | 213 (8.69) | |
| 6 | 903 (1.88) | 87 (3.55) | |
| 7 | 322 (0.67) | 38 (1.55) | |
| 8 | 74 (0.15) | 5 (0.20) | |
| 9 | 8 (0.02) | 2 (0.08) | |
| CHA2DS2-VASc score | 1.84±1.56 | 2.85±1.55 | <0.001‡ |
| Follow-up (years) | 6.91±3.51 | 4.08±3.00 | <0.001‡ |
Data given as n (%) or mean±SD. †Chi-squared test; ‡t-test. AF, atrial fibrillation; CHF, congestive heart failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; CVA, cerebrovascular accident; TIA, transient ischemic attack.
The incidence of new-onset AF increased from 1.27% in COPD patients with CHA2DS2-VASc score 0 to 9.17% in COPD patients with CHA2DS2-VASc score ≥6 (Figure 2). After adjusting for age and sex, as compared with COPD patients with CHA2DS2-VASc score 0, the corresponding adjusted hazard ratios (aHR) for new-onset AF increased with increasing CHA2DS2-VASc score. Furthermore, after adjusting for cancer, hyperlipidemia, chronic kidney disease, hyperthyroidism and valvular heart disease, as compared with COPD patients with CHA2DS2-VASc score 0, the corresponding aHR for new-onset AF for CHA2DS2-VASc score 1, 2, 3, 4, 5 and ≥6, were 1.24 (95% confidence interval (CI): 1.01–1.52), 1.56 (95% CI: 1.26–1.93), 1.72 (95% CI: 1.38–2.15), 1.80 (95% CI: 1.42–2.28), 2.14 (95% CI: 1.73–2.90) and 2.15 (95% CI: 1.62–2.86), respectively (trend test, P<0.001; Table 2).
Incidence of new-onset atrial fibrillation (AF) increased with increasing CHA2DS2-VASc score in chronic obstructive pulmonary disease patients.
| Variables | n | No. events |
Person- years |
Rate per 1,000 person-years |
Age- and sex- adjusted HR |
95% CI | Adjusted HR† |
95% CI |
|---|---|---|---|---|---|---|---|---|
| CHA2DS2-VASc score | ||||||||
| 0 | 9,760 | 124 | 73,153 | 1.70 | 1 | (Ref.) | 1 | (Ref.) |
| 1 | 15,015 | 354 | 111,981 | 3.16 | 1.45 | (1.17–1.79)*** | 1.24 | (1.01–1.52)* |
| 2 | 10,314 | 583 | 71,701 | 8.13 | 1.89 | (1.52–2.35)*** | 1.56 | (1.26–1.93)*** |
| 3 | 7,406 | 625 | 46,050 | 13.6 | 2.26 | (1.79–2.85)*** | 1.72 | (1.38–2.15)*** |
| 4 | 4,417 | 419 | 23,971 | 17.5 | 2.60 | (2.02–3.35)*** | 1.80 | (1.42–2.28)*** |
| 5 | 2,079 | 213 | 9,406 | 22.7 | 3.14 | (2.39–4.12)*** | 2.14 | (1.73–2.90)*** |
| ≥6 | 1,439 | 132 | 5,103 | 25.9 | 3.28 | (2.43–4.43)*** | 2.15 | (1.62–2.86)*** |
| P for trend | <0.001 | <0.001 | ||||||
*P<0.05, **P<0.01, ***P<0.001.† Adjusted for cancer, hyperlipidemia, CKD, hyperthyroidism and valvular heart disease. Abbreviations as in Table 1.
Figure 3 shows the 12-year survival probability curves for AF stratified by CHA2DS2-VASc score. There was a close approximation between the estimated and actual curves, suggesting that the CHA2DS2-VASc score did not overfit the data in this cohort.
Actual vs. estimated survival curves for new-onset atrial fibrillation stratified by CHA2DS2-VASc score in patients with chronic obstructive pulmonary disease.
The AUC for CHA2DS2-VASc score in predicting new-onset AF in patients with COPD was 0.69 (95% CI: 0.68–0.70; Figure 4). The Harrell C-statistic for incident AF was 0.73 based on Cox-regression analysis. The results of the univariable and multivariable Cox proportional hazards regression models for analyzing the risk of new-onset AF contributed by the factors of CHA2DS2-VASc score are listed in Table 3. Compared with patients aged ≤64 years, the risk of AF was 3.08-fold higher in those aged 65–74 years (95% CI: 2.76–3.44), and 5.64-fold higher in those aged ≥75 years (95% CI: 5.04–6.30). The aHR of AF was 1.21-fold higher in men than in women (95% CI: 1.12–1.31). The risk of AF was greater in patients with comorbidities, namely congestive heart failure (aHR, 2.67, 95% CI: 2.39–2.98) and hypertension (aHR, 1.76, 95% CI: 1.59–1.94).
Receiver operating characteristic (ROC) curve for CHA2DS2-VASc score in predicting new-onset atrial fibrillation in patients with chronic obstructive pulmonary disease (area under the curve, 0.6884).
| Variable | Crude | Adjusted† | ||
|---|---|---|---|---|
| HR | (95% CI) | HR | (95% CI) | |
| Age (years) | ||||
| ≤64 | 1.00 | (Ref.) | 1.00 | (Ref.) |
| 65–74 | 4.10 | (3.69–4.55)*** | 3.08 | (2.76–3.44)*** |
| ≥75 | 8.11 | (7.31–8.99)*** | 5.64 | (5.04–6.30)*** |
| Sex | ||||
| Male | 1.25 | (1.15–1.35)*** | 1.21 | (1.12–1.31)*** |
| Female | 1.00 | (Ref.) | 1.00 | (Ref.) |
| Underlying disease (components of CHA2DS2-VASc score) | ||||
| CHF | 4.89 | (4.39–5.44)*** | 2.67 | (2.39–2.98)*** |
| Diabetes mellitus | 1.34 | (1.19–1.50)*** | 0.90 | (0.80–1.01) |
| CVA or TIA | 2.30 | (2.01–2.63)*** | 1.13 | (0.98–1.30) |
| Vascular disease | 1.92 | (1.64–2.24)*** | 1.09 | (0.93–1.27) |
| Hypertension | 3.27 | (2.99–3.58)*** | 1.76 | (1.59–1.94)*** |
***P<0.001. †Adjusted for cancer, hyperlipidemia, chronic kidney disease, hyperthyroidism and valvular heart disease. Abbreviations as in Table 1.
To the best of our knowledge, the present study is the first to investigate the accuracy of CHA2DS2-VASc score in predicting new-onset AF in COPD patients. New-onset AF risk was found to be increased in parallel with increasing CHA2DS2-VASc score.
Coexistence of COPD and AF is common and the interplay between them is complex. COPD patients are at an increased risk of incident AF.28–30 Additionally, the adverse impact of incident AF on the clinical outcomes in COPD has also been proposed.12–14 To our knowledge, however, there is no research specifically addressing the risk prediction model for incident AF in COPD. In order to fill the knowledge gaps, the present study investigated the role of CHA2DS2-VASc score in predicting incidence of AF, using a large-scale cohort of 50,430 COPD patients and a mean follow-up period of 6.77 years. CHA2DS2-VASc score was found to be predictive of incident AF in COPD patients. More studies are needed to clarify the present findings.
Although several risk stratification models for AF development have been proposed,31,32 to the best of our knowledge, research on AF prediction modeling for patients with COPD is currently unavailable. In the present study, risk of new-onset AF was increased in COPD patients with CHA2DS2-VASc score ≥1, compared with CHA2DS2-VASc score 0. Unlike other complicated risk stratification models for AF prediction, the advantages of using CHA2DS2-VASc score in predicting incident AF are its convenience, comprehensiveness and accuracy.31,32 CHA2DS2-VASc score was also found to have moderate discriminative performance. Thus, CHA2DS2-VASc score is potentially useful in clinical settings.
Study LimitationsFirst, underestimation of AF incidence in this nationwide dataset is possible because AF cannot be identified if the patients do not seek medical advice. Second, personal health-associated behaviors were not provided by this nationwide database. Third, the severity of COPD and the types of AF were not available. Finally, we were unable to compare the predictive performance of the previously reported AF risk scores31,32 with the established CHA2DS2-VASc score due to the limitations of this nationwide database.
The current study is the first to evaluate the accuracy of CHA2DS2-VASc score in predicting incident AF in patients with COPD. The discriminatory capacity of the score was moderate. Therefore, CHA2DS2-VASc score might provide an easy way to stratify risk of new-onset AF in COPD patients.
This study was supported in part by Taiwan Ministry of Health and Welfare Clinical Trial and Research Center of Excellence (MOHW105-TDU-B-212-133019), China Medical University Hospital, Academia Sinica Taiwan Biobank Stroke Biosignature Project (BM10501010037), NRPB Stroke Clinical Trial Consortium (MOST 104-2325-B-039 -005), Tseng-Lien Lin Foundation, Taichung, Taiwan, Taiwan Brain Disease Foundation, Taipei, Taiwan, and Katsuzo and Kiyo Aoshima Memorial Funds, Japan.
The authors declare no conflict of interest.
