The CHADS 2 score is a validated clinical tool used for the risk stratification of stroke in the presence of atrial fibrillation (AF). Recently, some studies have shown that CHADS 2 score may predict the risk of AF, which yielded conflicting results. The purpose of this study is to perform a meta-analysis of observational studies to examine the association between the CHADS 2 score and risk of AF. Using PubMed and EMBASE database, we searched published articles by November 2014 to identify studies that evaluated the association between CHADS 2 score and the risk of AF. We used both fixed-effects and random-effects models to calculate the overall effect estimate. A sensitivity analysis and subgroup analysis were performed to find the origin of heterogeneity. Of the 1,806 studies identified initially, 19 studies were included into our analysis, with a total of 714,672 patients. The CHADS 2 score was found to be an independent predictor of AF as both a continuous variable (odds ratio 1.43, 95% confidence interval 1.10 to 1.86, p = 0.007) and categorical variable (odds ratio 3.37, 95% confidence interval 2.65 to 4.28, p <0.00001). Subgroup analysis revealed that different patients’ age in study population may be a possible reason for the significant heterogeneity in our meta-analysis. In conclusion, CHADS 2 score predicts the risk of AF. Addressing risk factors and early recognition of AF are important and also awareness of CHADS 2 score to reduce stroke risk with pharmacologic prophylaxis.
The CHADS 2 score has been well validated and widely used for risk stratification of stroke in patients with atrial fibrillation (AF). This score system evaluates the risk of stroke through the sum of individual risk factors for stroke, including heart failure, hypertension, age ≥75 years, diabetes mellitus, and previous stroke or transient ischemic attack [doubled]). Since the development of CHADS 2 score, there have been numerous studies on using the score to predict development of AF. These studies have yielded conflicting results. We conducted a comprehensive meta-analysis to evaluate the present evidence and investigate whether the higher CHADS 2 score increases the occurrence of AF.
Methods
Two reviewers systematically and independently searched PubMed and EMBASE using the key terms “CHADS 2 score” and “atrial fibrillation” for studies published by November 2014. The reviewers examined the titles, abstracts, and reference lists of all articles to identify all potentially relevant studies. Additionally, a manual search was conducted using review articles on this topic and bibliographies of original articles. Finally, congress abstracts during the past 3 years were also searched manually.
Studies were included if they met the following criteria: (1) the study design was a prospective cohort study, retrospective cohort study, or case-control study; (2) inclusion of CHADS 2 score at baseline; (3) clearly defined end point events of the patients; (4) the odds ratio (OR) or hazard ratio (HR) of AF incidence and the corresponding 95% confidence interval (CI) for CHADS 2 score were reported; and (5) only studies that included patients with different CHADS 2 scores that is clearly defined and in accordance with current guideline–based definitions were selected. We included published and unpublished studies without language restriction.
Two investigators reviewed the abstracts to identify all eligible studies. Relevant studies were retrieved as full text and assessed for compliance with the inclusion criteria. Any discrepancies were resolved through consensus of a third reviewer.
Two blinded reviewers independently performed data extraction using a standard data extraction form. We extracted and analyzed all the multivariate adjusted OR/HR and the corresponding 95% CI. Adjusted OR values were selected for the analysis. The main potential confounders used for adjustments are body mass index, dyslipidemia, smoking, women, anemia, elevated creatinine, dialysis, peripheral vascular disease, and valvular heart disease. The extracted data of this study included first author’s last name, publication year, study design, sample size, AF definition, age and gender, follow-up duration, adjusted variables, and end point events.
To limit the heterogeneity secondary to differences among study designs, the quality of each study was evaluated according to the guidelines developed by the US Preventive Task Force and the Evidence-Based Medicine Working Group and was evaluated according to the 10-item Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist. The following characteristics were assessed: (1) clear description of inclusion and exclusion criteria, (2) study sample representative for mentioned population, (3) clear description of sample selection, (4) full specification of clinical and demographic variables, (5) sufficient duration of follow-up, (6) reports on loss of follow-up, (7) clear definition of AF, (8) clear definition of outcomes and outcome assessment, (9) temporality (assessment of CHADS 2 at the baseline), and (10) adjustment of possible confounders on the multivariate analysis. If a study did not clearly mention one of these key points, we considered that it had not been performed. Therefore, the reported characteristics may be underestimated.
Pooled effect sizes were presented as the OR values with 95% CI. The HR values in multivariate Cox proportional hazards model in each primary study were directly considered as OR values. As the studies included in this meta-analysis used CHADS 2 score as either a continuous or categorical, we performed a separate meta-analysis for both types of variable to evaluate the association between CHADS 2 score and the occurrence of AF. To evaluate the heterogeneity across studies, we used I 2 derived from the chi-square test, which describes the percentage of the variability in effect estimates resulting from heterogeneity, rather than sampling error. An I 2 >50% indicates at least moderate statistical heterogeneity. When pooled analysis resulted in significant heterogeneity, the random-effects model was used. We conducted fixed-effects meta-analysis using the inverse variance method for pooling effect sizes and random-effects meta-analysis using the inverse variance heterogeneity method. The sensitivity analysis was also done in a random predefined manner. We also performed separate analyses based on age (>60 or <60 years), different study population (the patients who received catheter ablation, those after myocardial infarction, those who had cardiac surgery, and others), and AF type (new-onset AF or recurrent AF). Publication bias was evaluated using the funnel plot. Statistical significance was defined as a 2-tailed p value of 0.05. All statistical analyses were performed with Review Manager, version 5.3 (Revman; The Cochrane Collaboration, Oxford, United Kingdom).
Results
A flow diagram of the data search and study selection is presented in Figure 1 . A total of 1,806 studies were found using our search criteria. We discarded 456 duplicate studies. Of the remaining 1,350 studies, we excluded 1,322 studies because they were either unrelated, irrelevant, review articles, or animal studies. The remaining 28 studies were then retrieved for detailed evaluation. Of these 28 studies, we excluded 9 studies because 7 had no OR/HR values in detail, 1 had no clearly defined end point, and 1 did not provide 95% confidence interval. The remaining 19 studies were with a total of 714,672 patients, of which 11,539 with and 703,133 without AF were included into our meta-analysis. The follow-up periods varied from 5 days to 9 years. The characteristics and quality assessments of each study are listed in Table 1 , and the demographics of patients in each study are listed in Table 2 . Of the 19 selected studies, 16 demonstrated that a higher CHADS 2 score is associated with either the risk of new-onset AF or AF recurrence, and the remaining 3 showed no significant association between CHADS 2 score and risk of AF after multivariate analysis; 12 studies with OR/HR values were obtained in which CHADS 2 score was analyzed as a continuous variable. Nine studies in which CHADS 2 score was analyzed as a categorical variable had been included in a separate analysis. Of the 19 included studies, 2 studies analyzed CHADS 2 score as both a continuous variable and categorical variable.
| Investigator (year) | Location | Number of patients | Study population | Study design | Mean follow-up (months) | Endpoint | AF incidence | Quality score | Mean CHADS 2 score |
|---|---|---|---|---|---|---|---|---|---|
| Barnett 2012 | USA | 1153 | All pacemaker patients without prior history of atrial tachyarrhythmias. | Cohort | 11.6 | New-onset AF | 12.6% | 7 | 1.69 |
| Canpolat 2014 | Turkey | 363 | Patients who underwent initial pulmonary vein isolation with cryoballoon technique for documented AF. | Cohort | 19.2±6.1 | Recurrence of AF. | 18.7% | 7 | 1.52±0.81 |
| Chao (1) 2013 | Taiwan | 702502 | Patients ≥age 18 and had no past history of cardiac arrhythmias or rheumatic heart disease. | Cohort | 108±26.4 | New-onset AF. | 1.3% | 7 | 0.17 |
| Chao (2) 2012 | Taiwan | 238 | Symptomatic drug refractory AF who received radiofrequency catheter ablation. | Cohort | 60 | Recurrence of AF. | 43.7% | 8 | NA |
| Chao (3) 2011 | Taiwan | 247 | Patients with symptomatic drug-refractory paroxysmal AF who received catheter ablation. | Cohort | 17.3±7.0 | Recurrence of AF. | 23.1% | 8 | 0.98 |
| Chua 2013 | Taiwan | 277 | Patients who underwent cardiac surgery. | Cohort | 30 days | Postoperative AF. | 30% | 7 | 1.94 |
| Hu 2014 | China | 227 | AF patients who received single catheter ablation. | Cohort | 51 | Recurrence of AF. | 48.0% | 8 | NA |
| Huang 2013 | Taiwan | 724 | Patients with AMI. | Cohort | During hospitalization | New-onset AF | 10.8% | 7 | 1.6 |
| Kimura 2014 | Japan | 44 | AF patients who underwent an initial catheter ablation. | Cohort | 9.7±2.4 | Recurrence of AF. | 34.1% | 8 | 1.1±1.1 |
| Kornej 2014 | Germany | 2069 | Patients undergoing AF catheter ablation. | Cohort | ≥12 months after catheter ablation | Recurrence of AF. | 31.8% | 9 | 1.2±0.9 |
| Lau 2014 | China | 607 | Post-STEMI patients with no previously documented AF. | Cohort | 63±44 | New-onset AF. | 13.7% | 8 | 1.28 |
| Letsas 2014 | Greece | 126 | Patients with symptomatic, drug-refractory paroxysmal AF who underwent left atrial ablation. | Cohort | 16 | Recurrence of AF. | 29.4% | 9 | 0.93 |
| Manna 2013 | Italy | 304 | Kidney or kidney/liver transplant recipients. | Case control | During hospitalization | Postoperative AF | 8.2% | 7 | NA |
| Nagarajan 2011 | USA | 654 | Patients who underwent radiofrequency ablation for AF. | Cohort | 12 | Recurrence of AF. | 12.1% | 7 | NA |
| Richards 2013 | USA | 1157 | Pacemaker patients without a prior history of AF. | Cohort | 21.5 ± 4.7 | New-onset AF. | 21.0% | 8 | 1.9 ± 1.2 |
| Ruwald 2014 | Finland | 297 | Post-MI patients who received an implantable cardiac monitor. | Cohort | 23 | AF, bradyarrhythmias, ventricular tachycardia and ventricular fibrillation; major cardiovascular events. | 34.0% | 9 | NA |
| Sareh 2014 | USA | 2120 | Patients who underwent cardiac surgery. | Case control | 9.4±10.8 days | Postoperative AF | 16.2% | 8 | 1.8±1.3 |
| Zhang 2014 | China | 1035 | Patients with AMI(including STEMI and non- STEMI ). | Case control | During hospitalization | New-onset AF. | 7.44% | 8 | 1.50 |
| Zuo 2013 | China | 528 | Patients for assessment of palpitation, dizziness, and/or syncope. | Cohort | 73 | New-onset AF. | 16.8% | 9 | 1.3±1.3 |
| Investigator (year) | Mean age (years) | Male | HTN | HF | DM | Prior TIA / stroke | Mean LAD(mm) | Mean LVEF (%) | Medications | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| β-blocker | ACEI /ARB | Statin | Aspirin | Warfarin | |||||||||
| Barnett 2012 | 73.0±12.0 | 56.4% | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA |
| Canpolat 2014 | 53.5±11.2 | 52.6% | 36.4% | NA | 12.7% | NA | 37.7±4.7 | 65.1±4.2 | NA | 35.2% | 18.7% | NA | NA |
| Chao (1) 2013 | 41.3±16.4 | 50.9% | 5.2% | 0.4% | 3.1% | 1.8% | NA | NA | NA | NA | NA | NA | NA |
| Chao (2) 2012 | 53.2±12.8 | 71.4% | 53.8% | 7.6% | 18.1% | 6.7% | 38.3±5.9 | 54.5±6.7 | NA | NA | NA | NA | NA |
| Chao (3) 2011 | 52.8±12.8 | 72.1% | 38.1% | 11.3% | 25.1% | 8.9% | 38.4 | 57.8 | NA | 18.6% | NA | NA | NA |
| Chua 2013 | 62.1±9.7 | 76.9% | 76.9% | 24.2% | 55.2% | 13.7% | 39.8 | 58.8 | 71.8% | 60.3% | 54.9% | 56.7% | NA |
| Hu 2014 | 60.7±10.9 | 58.1% | 58.2% | 11.9% | 10.1% | 8.8% | 39.34 | 66.8 | NA | NA | NA | NA | NA |
| Huang 2013 | 67±12 | 80% | 64.1% | 5.9% | 36.6% | 12.2% | 39.5 | 44.0 | 52.6% | 59.7% | 37.7% | NA | NA |
| Kimura 2014 | 59 ± 8 | 86.4% | 52.3% | 15.9% | 18.2% | 9.1% | 39±6 | 71.9±9.4 | 45.5% | 38.6% | 22.7% | 2.3% | 68.2% |
| Kornej 2014 | 60±10 | 66% | 71% | 7% | 15% | 9% | 43±6 | 59±10 | NA | NA | NA | NA | NA |
| Lau 2014 | 64.7 | 75.1% | 52.2% | 14.7% | 38.9% | NA | NA | 45.6 | 72.8% | 79.1% | 8.1% | 93.7% | NA |
| Letsas 2014 | 61 | 61.9% | 44.4% | NA | 11.9% | NA | NA | 65 | 13.5% | 38.9% | 32.1% | NA | NA |
| Manna 2013 | 51.1±12.5 | 67.1% | 60.2% | NA | 5.3% | 3.0% | 39.25 | NA | NA | NA | NA | NA | NA |
| Nagarajan 2011 | 60.9 | 72.5% | NA | NA | NA | NA | 42.3 | 57.1 | NA | NA | NA | NA | NA |
| Richards 2013 | 73.0 ± 12.0 | 56.4% | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA |
| Ruwald 2014 | 64.0 | 77.1% | 43.4% | 76.4% | 19.9% | 10.4% | NA | 31.6 | 96.3% | 90.2% | 82.5% | 90.6% | NA |
| Sareh 2014 | 62.1±14.7 | 66.2% | 66.2% | 43.4% | 26.7% | NA | NA | NA | 60.2% | NA | 58.6% | 51.2% | 4.6% |
| Zhang 2014 | 65.22±12.14 | 67.0% | 51.9% | 25.7% | 23.8% | 12.3% | 37.0 | 52.4 | 52.9% | 74.0% | 77.7% | NA | NA |
| Zuo 2013 | 68.5±10.6 | 46.2% | 50.9% | 5.1% | 18.6% | 11.7% | 38±8 | 63±10.4 | 46.2% | 43.2% | NA | 41.1% | NA |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
