- Email: [email protected]
Atrial Flutter Patients
CHADS2 and CHA2DS2-VASc Risk Factors to Predict First Cardiovascular Hospitalization Among Atrial Fibrillation/Atrial Flutter Patients Gerald V. Naccarelli, MDa,*, Mary Prince Panaccio, PhDc, Gordon Cummins, MSb, and Nora Tu, PhDc Limited data exist concerning risk factors for cardiovascular (CV) hospitalization in patients with atrial fibrillation (AF) or atrial flutter (AFL). The aim of this retrospective cohort evaluation was to assess whether patient characteristics and risk factors, including CHADS2 (congestive heart failure, hypertension, age >75 years, type 2 diabetes, and previous stroke or transient ischemic attack [doubled]) and CHA2DS2-VASc (congestive heart failure; hypertension; age >75 years [doubled]; type 2 diabetes; previous stroke, transient ischemic attack, or thromboembolism [doubled]; vascular disease; age 65 to 75 years; and sex category) scores, identified patients with AF or AFL at risk for CV hospitalization. Claims data (January 2003 to June 2009) were evaluated to identify patients aged >40 years with >1 inpatient or >2 (within 30 days of each other) outpatient diagnoses of AF or AFL and an absence of diagnosis codes related to cardiac surgery within 30 days of AF or AFL diagnosis. Risk factors for first CV hospitalization in the 2-year period after diagnosis were assessed using univariate and multivariate analyses. Overall, 377,808 patients (mean age 73.9 ⴞ 12.1 years) were identified, of whom 128,048 had CV hospitalizations. CHADS2 and CHA2DS2-VASc scores were the top 2 predictors of first CV hospitalization after AF or AFL diagnosis. Hospitalization risk was increased 2.3- to 2.7-fold in patients with CHADS2 scores of 6 and approximately 3.0-fold in patients with CHA2DS2-VASc scores of 9 compared to patients with a score of 0. These increases were maintained essentially unchanged throughout the 2-year follow-up period. In conclusion, CHADS2 and CHA2DS2-VASc scores were predictive of first CV hospitalization in patients with AF or AFL and may be helpful in identifying “at-risk” patients and guiding therapy. © 2012 Elsevier Inc. All rights reserved. (Am J Cardiol 2012;109:1526 –1533) An estimated 3.03 million patients had atrial fibrillation (AF) in the United States in 2005, and this number is expected to reach 7.56 million by 2050.1 Patients with AF or atrial flutter (AFL) frequently have cardiovascular (CV) co-morbidities and are at increased risk for hospitalization.2– 4 AF or AFL is responsible for approximately 529,000 hospitalizations in the United States each year.5 Because the mean cost per CV hospitalization is approximately $10,908,6 these hospitalizations have a significant impact on the economic cost of caring for these patients.7 Validated schemas for assessing the risk for stroke, thromboembolism, and mortality in patients with AF or AFL are used to guide antithrombotic therapy and include the CHADS2 (congestive heart failure, hypertension, age ⱖ75 years, type 2 diabetes, and previous stroke or transient isch-
a Penn State Heart & Vascular Institute, Penn State Milton S. Hershey Medical Center, Hershey, Pennsylvania; bQuintiles, Hawthorne, New York; and cSanofi-Aventis U.S., Bridgewater, New Jersey. Manuscript received November 18, 2011; revised manuscript received and accepted January 12, 2012. This study was funded by Sanofi-Aventis U.S., Bridgewater, New Jersey. Editorial support for this report was provided by Peloton Advantage, LLC, Parsippany, New Jersey, funded by Sanofi-Aventis U.S. *Corresponding author: Tel: 717-531-3907; fax: 717-531-4077. E-mail address: [email protected] (G.V. Naccarelli).
0002-9149/12/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2012.01.371
emic attack [doubled]) and CHA2DS2-VASc (congestive heart failure; hypertension; age ⱖ75 years [doubled]; type 2 diabetes; previous stroke, transient ischemic attack, or thromboembolism [doubled]; vascular disease; age 65 to 75 years; and sex category) scores.8 –11 The objective of this evaluation was to determine whether specific patient characteristics or risk factors, including CHADS2 and CHA2DS2-VASc scores, were predictive of first CV hospitalization in patients with AF or AFL. Methods This retrospective cohort evaluation used integrated patient data for the period from January 1, 2003, to June 30, 2009, obtained from the Thomson Reuters (Cambridge, Massachusetts) MarketScan Commercial Claims and Encounters database and Medicare Supplemental (January 1, 2003, to December 31, 2008) database (Figure 1). This database contains detailed enrollment, clinical utilization, and expenditure data for approximately 85 million subjects from approximately 45 large employers, health plans, and government and public organizations. These data reflect real-world treatment and outcomes and have been used for ⬎400 peer-reviewed publications.12 In accordance with the Health Insurance Portability and Accountability Act of 1996, all patient data were deidentified before analysis. The study population included patients aged ⱖ40 years www.ajconline.org
Review/CHADS2/CHA2DS2-VASc and Hospitalization in AF
1527
MarketScan Enrollees 01/01/2003–6/30/2009 N=84,744,071
Inclusion Criteria • AF index date between 01/01/2004–6/29/2009 (Medicaid: 12/30/2008) o Included: n=1,121,251 (1.3%); Excluded: n=83,652,820 (98.7%) • 12 months continuous enrollment prior to AF index date o Included: n=559,300 (49.9%); Excluded: n=561,951 (50.1%) • ≥1 day of information following AF index date and continuous enrollment during the entire follow-up period o Included: n=500,612 (89.5%); Excluded: n=58,688 (10.5%) • ≥40 years of age on AF index date o Included: n=492,446 (98.4%); Excluded: n=8,166 (1.6%) Exclusion Criteria • Inpatient or outpatient procedure codes related to cardiac surgery (excluding codes related to pacemaker and ablation) ≤30 days prior to AF index date o Patients meeting exclusion criteria: n=114,638 (23.3%) Study Population N=377,808 Inpatient/ER AF patients: n=199,412 (52.8%) Outpatient AF patients: n=178,396 (47.2%)
Yes
1st CV Hospitalization?
n=128,048 (33.9%)
No
n=249,760 (66.1%)
Figure 1. Patient identification. ER ⫽ emergency room.
with diagnoses of AF or AFL (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9CM] diagnosis codes 427.31 and 427.32) during ⱖ1 inpatient admission or ⱖ2 outpatient medical claims from January 1, 2004, to June 29, 2009 (December 30, 2008, for Medicaid patients). The qualifying diagnoses could appear in any position on a claim for any inpatient admission or within 30 days of each other for outpatient service. The date of the first qualifying AF diagnosis from January 1, 2004, to June 29, 2009, was designated the index date. Data from all identified patients were then further evaluated and included in the analysis if the patient was aged ⱖ40 years on the index date, had ⱖ12 months of continuous enrollment before the index date, had ⱖ1 day of follow-up information after the index date, had continuous enrollment during the entire follow-up period, and did not have an ICD-9-CM code relating to cardiac surgery (excluding pacemaker and ablation codes) 30 days before the index date. The study consisted of baseline and follow-up periods. Data from the baseline period, which started 364 days before the index date and ended on the index date, were used to obtain information about each patient’s medical history. Data from the follow-up period, which started the day after
the index date and ended after 24 months of observation, when the patient exited the database, or on the study end date (whichever came first), were used to assess hospitalization risk. The primary outcome of interest was the first CV hospitalization after the index date with all-cause CV hospitalization defined by ICD-9-CM codes 390.xx to 459.xx. A secondary outcome measure was elapsed time from index date to first CV hospitalization (time to first hospitalization). Demographic and clinical characteristics at baseline were summarized using descriptive statistics (means, medians, standard deviations, and ranges for continuous variables and frequencies and percentages for categorical variables). Univariate and multivariate analyses using logistic regression were performed to identify predictors of CV hospitalization, defined as binary outcomes (CV hospitalization, yes or no). Logistic regression modeling13 was used as an exploratory tool to better understand the interrelations among many potentially correlated clinical factors by estimating the odds ratio (OR) and 95% confidence interval (CI). Multivariate Cox proportional-hazard models14 were used to estimate hazard ratios (HR) and 95% CIs, to evaluate time to first CV hospitalization, and to understand the
1528
The American Journal of Cardiology (www.ajconline.org)
Table 1 Atrial fibrillation cohort demographic and baseline characteristics at index event date, MarketScan claims data, January 1, 2003, to June 30, 2009 Characteristic
Demographics Age at index (years) Median (range) Mean ⫾ SD Age category (years) 40–74 ⱖ75 Men Continuous enrollment (years) Median (range) Mean ⫾ SD AF diagnosis at index Inpatient Outpatient/ER Type of AF diagnosis New case Existing case Region Northeast North Central South West Unknown* Plan type Commercial Medicare Medicaid Baseline characteristics Charlson co-morbidity index 0 ⱖ1 Previous CV hospitalization Hypertension Stroke Myocardial infarction Pulmonary embolism Major bleeding Structural heart disease Other coronary artery disease Heart failure Cancer Valvular heart disease Cardiac hypertrophy Nonischemic cardiomyopathy Chronic obstructive pulmonary disease Type 2 diabetes Treated hyperthyroidism Composite AF symptom† CHADS2 score 0 1 2 3 4 5 6 CHA2DS2-VASc score 0 1 2
All Patients (n ⫽ 377,808)
76 (40–109) 73.9 ⫾ 12.1 171,026 (45.3%) 206,782 (54.7%) 196,840 (52.1%) 3.5 (1.01–6.5) 4.1 ⫾ 1.8
Patients With First CV Hospitalization (n ⫽ 128,048)
77 (40–108) 74.9 ⫾ 11.5 53,140 (41.5%) 74,908 (58.5%) 66,179 (51.7%) 4.0 (1.03–6.5) 4.3 ⫾ 1.7
Patients With No CV Hospitalizations (n ⫽ 249,760)
75 (40–109) 73.4 ⫾ 12.4 117,886 (47.2%) 131,874 (52.8%) 130,661 (52.3%) 3.5 (1.01–6.5) 4.0 ⫾ 1.9
199,412 (52.8%) 178,396 (47.2%)
72,076 (56.3%) 55,972 (43.7%)
127,336 (51.0%) 122,424 (49.0%)
238,443 (63.1%) 139,365 (36.9%)
59,004 (46.1%) 69,044 (53.9%)
179,439 (71.8%) 70,321 (28.2%)
31,578 (8.4%) 113,261 (30.0%) 106,797 (28.3%) 87,889 (23.3%) 38,283 (10.1%)
10,854 (8.5%) 40,294 (31.5%) 34,758 (27.1%) 29,786 (23.3%) 12,356 (9.6%)
20,724 (8.3%) 72,967 (29.2%) 72,039 (28.8%) 58,103 (23.3%) 25,927 (10.4%)
85,689 (22.7%) 254,664 (67.4%) 37,455 (9.9%)
23,488 (18.3%) 92,471 (72.2%) 12,089 (9.4%)
62,201 (24.9%) 162,193 (64.9%) 25,366 (10.2%)
203,089 (53.8%) 174,719 (46.2%) 242,740 (64.2%) 218,073 (57.7%) 54,504 (14.4%) 20,310 (5.4%) 7,616 (2.0%) 70,682 (18.7%) 191,402 (50.7%) 141,741 (37.5%) 117,647 (31.1%) 113,449 (30.0%) 99,140 (26.2%) 50,047 (13.2%) 28,844 (7.6%) 77,053 (20.4%) 76,735 (20.3%) 30,686 (8.1%) 120,964 (32.0%)
59,499 (46.5%) 68,549 (53.5%) 89,611 (70.0%) 77,778 (60.7%) 20,682 (16.2%) 7,830 (6.1%) 2,666 (2.1%) 26,784 (20.9%) 74,469 (58.2%) 56,168 (43.9%) 49,404 (38.6%) 39,241 (30.6%) 37,402 (29.2%) 20,628 (16.1%) 12,254 (9.6%) 31,448 (24.6%) 30,584 (23.9%) 10,933 (8.5%) 44,110 (34.4%)
143,590 (57.5%) 106,170 (42.5%) 153,129 (61.3%) 140,295 (56.2%) 33,822 (13.5%) 12,480 (5.0%) 4,950 (2.0%) 43,898 (17.6%) 116,933 (46.8%) 85,573 (34.3%) 68,243 (27.3%) 74,208 (29.7%) 61,738 (24.7%) 29,419 (11.8%) 16,590 (6.6%) 45,605 (18.3%) 46,151 (18.5%) 19,753 (7.9%) 76,854 (30.8%)
50,615 (13.4%) 105,004 (27.8%) 109,002 (28.9%) 66,172 (17.5%) 31,139 (8.2%) 13,053 (3.5%) 2,823 (0.7%)
11,598 (9.1%) 31,539 (24.6%) 39,235 (30.6%) 26,607 (20.8%) 12,365 (9.7%) 5,462 (4.3%) 1,242 (1.0%)
39,017 (15.6%) 73,465 (29.4%) 69,767 (27.9%) 39,565 (15.8%) 18,774 (7.5%) 7,591 (3.0%) 1,581 (0.6%)
15,776 (4.2%) 34,867 (9.2%) 54,564 (14.4%)
3,051 (2.4%) 8,063 (6.3%) 15,738 (12.3%)
12,725 (5.1%) 26,804 (10.7%) 38,826 (15.5%)
Review/CHADS2/CHA2DS2-VASc and Hospitalization in AF
1529
Table 1 (Continued) All Patients (n ⫽ 377,808)
Patients With First CV Hospitalization (n ⫽ 128,048)
Patients With No CV Hospitalizations (n ⫽ 249,760)
75,261 (19.9%) 80,105 (21.2%) 59,419 (15.7%) 35,069 (9.3%) 15,959 (4.2%) 5,681 (1.5%) 1,107 (0.3%)
24,911 (19.5%) 29,215 (22.8%) 23,343 (18.2%) 14,179 (11.1%) 6,632 (5.2%) 2,408 (1.9%) 508 (0.4%)
50,350 (20.2%) 50,890 (20.4%) 36,076 (14.4%) 20,890 (8.4%) 9,327 (3.7%) 3,273 (1.3%) 599 (0.2%)
Characteristic 3 4 5 6 7 8 9
* Includes Medicaid population of 37,455. Includes chest pain, tachycardia, palpitations, dizziness, syncope, dyspnea, and fatigue at AF index date.
†
Table 2 Atrial fibrillation cohort treatments at baseline, MarketScan claims data, January 1, 2003, to June 30, 2009 Treatment at Baseline
Rhythm-control drugs Rate-control drugs Antithrombotic drugs Statins Antihypertensive drugs Electrical cardioversion Ablation (CPT code 93651) Pacemaker Other cardiac surgery
All Patients (n ⫽ 377,808)
Patients With First CV Hospitalization (n ⫽ 128,048)
Patients With No CV Hospitalizations (n ⫽ 249,760)
16,443 (4.3%) 106,481 (28.2%) 88,444 (23.4%) 145,865 (38.6%) 197,263 (52.2%) 16,007 (4.2%) 5,422 (1.4%) 31,295 (8.2%) 175,243 (46.3%)
6,186 (4.8%) 40,838 (31.9%) 34,914 (27.3%) 52,280 (40.8%) 73,643 (57.5%) 5,840 (4.6%) 1,923 (1.5%) 12,644 (9.9%) 60,988 (47.6%)
10,257 (4.1%) 65,643 (26.3%) 53,530 (21.4%) 93,585 (37.5%) 123,620 (49.5%) 10,167 (4.1%) 3,499 (1.4%) 18,651 (7.5%) 114,255 (45.7%)
CPT ⫽ Current Procedural Terminology.
time dependency of various covariates (co-morbidities, treatments, demographics). All multivariate models used a backward selection method to remove statistically insignificant covariates, with a significance level ⱕ0.01 (the threshold required for any covariate to remain in the final model). Model fits were evaluated by Hosmer-Lemeshow test and C statistic.15 On the basis of clinical considerations, we reduced the number of initial variables (n ⫽ 36) to include the top 10 covariates. The life table method16 was used to estimate the percentage risk for hospitalization at different risk periods of 3, 6, 9, 12, 15, 18, 21, and 24 months for each respective CHADS2 and CHA2DS2-VASc score. For these models, risk was assessed at the end of each period. Results Of the nearly 85 million enrollees in the MarketScan database, 377,808 (0.45%) were identified who fulfilled all inclusion and exclusion criteria (Figure 1). Table 1 lists demographic and clinical characteristics at baseline for the overall study cohort and by CV hospitalization status. Differences between patients with first CV hospitalizations and no CV hospitalizations were statistically significant (p ⬍0.05) for all characteristics. Baseline CHADS2 and CHA2DS2-VASc factors were the strongest predictors of first CV hospitalization in patients with AF or AFL in all univariate and multivariate logistic and Cox proportional-hazards models. AF treatments and certain concomitant medications (listed in Table 2) were not
found to be predictive of first hospitalization. Therefore, models that included key co-morbidities provided the best clinical and statistical predictive results. On the basis of multivariate logistic models (age modeled as a continuous variable), congestive heart failure (OR 6.6, 95% CI 6.01 to 7.32), type 2 diabetes (OR 2.4, 95% CI 2.17 to 2.73), stroke (OR 3.1, 95% CI 2.68 to 3.57), and hypertension (OR 1.5, 95% CI 1.35 to 1.62) were found to increase the risk for first CV hospitalization (data not shown). In multivariate logistic regression models that included all variables (with age as a binary variable), the risk for first CV hospitalization was 3 times higher in patients with preexisting AF during baseline compared to patients with newly diagnosed AF. Therefore, estimates of time to first CV hospitalization were stratified by this variable. The individual components of CHADS2 (congestive heart failure: HR 1.7, 95% CI 1.65 to 1.71; type 2 diabetes: HR 1.3, 95% CI 1.28 to 1.33; stroke: HR 1.2, 95% CI 1.17 to 1.22; hypertension: HR 1.2, 95% CI 1.20 to 1.24) were also found to be associated with time to first CV hospitalization risk. However, because of violations in the underlying assumption that the HR is independent of time and valid only for time-independent covariates for the Cox proportional-hazards models, a nonparametric life table approach was taken to estimate the association of CHADS2 and CHA2DS2-VASc scores with first CV hospitalization risk. The correlation between CHADS2 and CHA2DS2-VASc
1530
The American Journal of Cardiology (www.ajconline.org)
higher) for scores ⬍2, but at higher scores, the differences in risk between these patients became more similar. In other words, the risk for CV hospitalization was similar between patients with new AF and those with preexisting AF if their baseline CHADS2 or CHA2DS2-VASc scores were already high. Discussion
Figure 2. Time to first CV hospitalization by CHADS2 (A) and CHA2DS2VASc (B) scores.
scores and risk for first CV hospitalization was present and relatively consistent from the first through the last month of the follow-up period (Figure 2). Relative to a CHADS2 score of 0, a score of 6 was associated with an excess 2.7-fold risk at 6 or 12 months, 2.5-fold risk at 18 months, and 2.3-fold risk at 24 months. A CHA2DS2-VASc score of 9 was associated with an approximate 3.0-fold increase in risk across time periods (6, 12, 18, and 24 months) compared to a score of 0 (Table 3). Patients who entered the cohort with preexisting AF had higher risks (OR 3.0, 95% CI 2.93 to 3.08) for CV hospitalization, even after adjustment for all other factors (treatment, age, other baseline co-morbidities listed in Table 1). For CHADS2 and CHA2DS2-VASc, the risk for CV hospitalization differed by whether patients were newly diagnosed or had preexisting diagnoses of AF at study entry (Figures 3 and 4). Relative to patients with preexisting diagnoses, the absolute risk for first CV hospitalization in newly diagnosed patients increased as CHADS2 and CHA2DS2-VASc scores increased, although this increase was constant over time. The differences in CV hospitalization risk between new and preexisting AF cases within a score category were greatest (relative risk about twofold
Approximately 4% of individuals aged ⱖ60 years in the United States have AF.17 This arrhythmia significantly increases the risk for stroke, other thromboembolic events, heart failure, hospitalization, and death.18 Approximately 31%, 48%, and 59% of patients with AF will be hospitalized for CV reasons (predominantly acute coronary syndromes, decompensated heart failure, thromboembolic events, and arrhythmia management) within 1, 3, and 5 years of diagnosis, respectively.4,18,19 These hospitalizations are the single largest contributor to the economic cost of AF.19 –21 Data indicate that the costs related to hospitalization in patients with AF equal or exceed those for all other medical care of these patients combined.7,19 –21 In a retrospective, claims-based analysis, annual direct medical costs for patients with and without AF were $15,553 and $2,792, respectively, with inpatient hospital costs ($8,486) accounting for 55% of the total direct cost of care in patients with AF.21 Similarly, in another retrospective, claims-based analysis, mean annual total health care costs for patients with and without AF were $23,750 and $7,439, respectively, with hospitalization costs ($13,507) accounting for 57% of the total health care costs in patients with AF.7 Risk factors for stroke, thromboembolism, and death have been extensively evaluated in patients with AF, and these evaluations have led to the development and validation of the CHADS2 and CHA2DS2-VASc scoring systems in an attempt to distinguish patients who might benefit from more intense antithrombotic therapy.8 –11 In contrast, only limited data are available to help identify patients with AF who are at high risk for hospitalization. A retrospective, community-based evaluation of approximately 4,500 individuals in Olmsted County, Minnesota, identified age, body mass index, systolic blood pressure, paroxysmal AF, and histories of myocardial infarction, valvular heart disease, peripheral artery disease, carotid artery disease, stroke, systemic hypertension, diabetes mellitus, chronic renal disease, and chronic obstructive pulmonary disease as independent risk factors for hospitalization in patients with AF.4 The present retrospective cohort evaluation is the largest and most geographically diverse assessment of hospitalization risk in patients with AF or AFL performed to date. In this assessment, CHADS2 and CHA2DS2-VASc scores were the 2 single best predictors of hospitalization risk in logistic regression and Cox proportional-hazards models. A CHADS2 score of 6 was associated with an approximate 2.5-fold increase, and a CHA2DS2-VASc score of 9 was associated with an approximate 3.0-fold increase in hospitalization risk relative to a score of 0, with comparatively proportional increments in risk for intermediate scores. Moreover, the predictive value of these scores was consistent throughout the evaluation period. Differences in CV
Review/CHADS2/CHA2DS2-VASc and Hospitalization in AF
1531
Table 3 Risk and 95% confidence intervals of first cardiovascular hospitalization by CHADS2 and CHA2DS2-VASc scores over time Score
CHADS2 0 1 2 3 4 5 6 CHA2DS2-VASc 0 1 2 3 4 5 6 7 8 9
% Risk (95% CI) 6 Months
12 Months
18 Months
24 Months
13.9 (13.58–14.21) 17.7 (17.47–17.96) 22.3 (22.05–22.59) 27.8 (27.42–28.18) 28.3 (27.73–28.85) 32.3 (31.42–33.26) 37.0 (34.92–39.13)
20.8 (20.42–21.20) 27.9 (27.63–28.24) 35.7 (35.34–36.00) 43.2 (42.75–43.66) 44.0 (43.34–44.68) 49.5 (48.43–50.59) 56.6 (54.25–59.06)
26.4 (25.98–26.88) 35.7 (35.35–36.05) 45.1 (44.77–45.51) 53.1 (52.62–53.60) 53.8 (53.03–54.49) 59.9 (58.80–61.10) 66.9 (64.42–69.39)
32.0 (31.47–32.55) 43.0 (42.58–43.40) 53.2 (52.80–53.65) 61.7 (61.11–62.22) 62.4 (61.52–63.18) 68.8 (67.50–70.14) 74.4 (71.51–77.12)
12.5 (11.95–13.03) 15.0 (14.63–15.42) 17.5 (17.19–17.87) 19.8 (19.49–20.10) 22.5 (22.22–22.84) 26.0 (25.61–26.39) 28.3 (27.75–28.81) 31.1 (30.26–31.90) 33.4 (32.00–34.83) 38.6 (35.33–42.08)
18.3 (17.62–18.95) 21.9 (21.37–22.34) 26.7 (26.24–27.07) 31.3 (30.88–31.64) 36.1 (35.69–36.46) 41.4 (40.90–41.84) 44.5 (43.89–45.16) 48.4 (47.47–49.40) 51.2 (49.61–52.91) 58.0 (54.26–61.80)
22.7 (21.98–23.53) 27.1 (26.60–27.71) 33.9 (33.45–34.40) 39.7 (39.30–40.14) 45.5 (45.08–45.94) 51.5 (51.03–52.06) 54.8 (54.10–55.48) 58.6 (57.52–59.59) 62.0 (60.25–63.74) 68.8 (64.93–72.67)
27.0 (26.13–27.99) 32.3 (31.62–32.93) 40.7 (40.18–41.30) 47.5 (47.02–48.00) 54.0 (53.46–54.45) 59.9 (59.27–60.44) 63.9 (63.12–64.70) 67.4 (66.21–68.55) 69.5 (67.48–71.42) 76.1 (71.54–80.37)
Figure 3. Time to first CV hospitalization by CHADS2 score stratified by preexisting (A) or new (B) AF diagnosis at index date.
Figure 4. Time to first CV hospitalization by CHA2DS2-VASc score stratified by preexisting (A) or new (B) AF diagnosis at index date.
1532
The American Journal of Cardiology (www.ajconline.org)
hospitalization risks with increasing scores was observed by type of patient diagnosis (those with preexisting AF diagnoses were at greater risk than those newly diagnosed with AF). Although the C statistics in these models (0.57 to 0.58) indicate only modest strength for predicting CV hospitalization risk, these statistics are similar to those observed for the more established use of these scoring systems in predicting thromboembolism risk.11 One possible explanation for the paucity of studies evaluating hospitalization risk in patients with AF is that until recently, there was little evidence that treatment could ameliorate this risk. The advantages of sinus rhythm in early clinical trials of AF therapy appear to have been offset by the deleterious effects of the antiarrhythmic drugs used, and hospitalization was more frequent in the rhythm-control arms of these trials.9 However, recent data from A PlaceboControlled, Double-Blind, Parallel Arm Trial to Assess the Efficacy of Dronedarone 400 mg b.i.d. for the Prevention of CV Hospitalization or Death from any Cause in Patients With Atrial Fibrillation/Atrial Flutter (ATHENA) demonstrated that treatment can safely and effectively reduce CV hospitalization risk.22 As a result, American and European treatment guidelines have recently been amended to recommend consideration of dronedarone use in patients with nonpermanent AF to reduce the risk for CV hospitalizations.18,23 The reduction in CV hospitalization with dronedarone appears to be limited to the ATHENA population, as patients with decompensated heart failure who met Antiarrhythmic Trial With Dronedarone in Moderate-to-Severe CHF Evaluating Morbidity Decrease (ANDROMEDA) inclusion criteria were excluded.22,24 Furthermore, the Permanent Atrial Fibrillation Outcome Study Using Dronedarone on Top of Standard Therapy (PALLAS) trial, designed to assess the efficacy of dronedarone compared to placebo in patients with permanent AF, was terminated prematurely because of a significant increase in CV events in the dronedarone arm.25 Further studies assessing the benefits of pharmacologic and nonpharmacologic rhythm control strategies in patients with AF are needed. Finally, it is important to consider limitations of the present evaluation. The evaluation was performed using information obtained from an administrative claims database. These data are typically collected to document justification for billing, not medical research, and thus clinical details, such as the type and severity of valvular disease, are often limited or unavailable. In addition, by relying on ICD-9-CM codes and National Drug Codes to define the study cohort, co-morbidities, drugs, and hospitalization, our analysis is vulnerable to inaccurate and/or differential coding practices of the sources providing information to the database. However, we used validated algorithms from the published research to minimize misclassification of AF; these algorithms have been found in multiple studies to have a high positive predictive value.26 –28 To further enhance specificity, we required ⱖ2 outpatient AF diagnoses occurring on different days or 1 inpatient diagnosis. Separating encounters by a minimal period (e.g., 30 days) has been used to increase specificity and identify chronic AF.29 Last, because of the large sample sizes in this analysis, small differences in demographics and baseline characteristics between patients with CV hospitalizations and those with no
CV hospitalizations resulted in significant p values that are clinically insignificant. Even covariates with ORs or HRs close to 1 had statistically significant p values because of the large sample size. Consequently, clinical and etiologic considerations guided selection of covariates and interpretation of modeling results. In summary, the risk for first CV hospitalization increased with time and CHADS2 and CHA2DS2-VASc scores in this retrospective cohort evaluation of patients with AF or AFL. As a result, this scoring system may be useful to clinicians in identifying patients at high risk for hospitalization who might benefit from targeted therapy to reduce this risk. Further investigation is warranted to validate these findings in other clinical and epidemiologic studies. Acknowledgment: We thank Gerald Barber, MD, for writing assistance. 1. Naccarelli GV, Varker H, Lin J, Schulman KL. Increasing prevalence of atrial fibrillation and flutter in the United States. Am J Cardiol 2009;104:1534 –1539. 2. Wattigney WA, Mensah GA, Croft JB. Increasing trends in hospitalization for atrial fibrillation in the United States 1985 through 1999: implications for primary prevention. Circulation 2003;108:711–716. 3. Miyasaka Y, Barnes ME, Gersh BJ, Cha SS, Bailey KR, Abhayaratna WP, Seward JB, Tsang TS. Secular trends in incidence of atrial fibrillation in Olmsted County, Minnesota 1980 to 2000, and implications on the projections for future prevalence. Circulation 2006;114: 119 –125. 4. Miyasaka Y, Barnes ME, Gersh BJ, Cha SS, Bailey KR, Seward JB, Tsang TS. Changing trends of hospital utilization in patients after their first episode of atrial fibrillation. Am J Cardiol 2008;102:568 –572. 5. Roger VL, Go AS, Lloyd-Jones DM, Adams RJ, Berry JD, Brown TM, Carnethon MR, Dai S, De Simone G, Ford ES, Fox CS, Fullerton HJ, Gillespie C, Greenlund KJ, Hailpern SM, Heit JA, Ho PM, Howard VJ, Kissela BM, Kittner SJ, Lackland DT, Lichtman JH, Lisabeth LD, Makuc DM, Marcus GM, Marelli A, Matchar DB, McDermott MM, Meigs JB, Moy CS, Mozaffarian D, Mussolino ME, Nichol G, Paynter NP, Rosamond WD, Sorlie PD, Stafford RS, Turan TN, Turner MB, Wong ND, Wylie-Rosett J. Heart disease and stroke statistics—2011 update: a report from the American Heart Association. Circulation 2011;123:459 – 463. 6. Naccarelli GV, Johnston SS, Lin J, Patel PP, Schulman KL. Cost burden of cardiovascular hospitalization and mortality in ATHENAlike patients with atrial fibrillation/atrial flutter in the United States. Clin Cardiol 2010;33:270 –279. 7. Lee WC, Lamas GA, Balu S, Spalding J, Wang Q, Pashos CL. Direct treatment cost of atrial fibrillation in the elderly American population: a Medicare perspective. J Med Econ 2008;11:281–298. 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. Fuster V, Ryden LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen KA, Halperin JL, Kay GN, Le Huezey JY, Lowe JE, Olsson SB, Prystowsky EN, Tamargo JL, Wann LS. 2011 ACCF/AHA/HRS focused updates incorporated into the ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines developed in partnership with the European Society of Cardiology and in collaboration with the European Heart Rhythm Association and the Heart Rhythm Society. J Am Coll Cardiol 2011;57:e101– e198. 10. Crandall MA, Horne BD, Day JD, Anderson JL, Muhlestein JB, Crandall BG, Weiss JP, Osborne JS, Lappe DL, Bunch TJ. Atrial fibrillation significantly increases total mortality and stroke risk be-
Review/CHADS2/CHA2DS2-VASc and Hospitalization in AF
11.
12.
13.
14.
15. 16.
17.
18.
yond that conveyed by the CHADS2 risk factors. Pacing Clin Electrophysiol 2009;32:981–986. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest 2010;137:263–272. Thompson Reuters. MarketScan studies, abbreviated bibliography. Available at: http://img.en25.com/Web/ThomsonReuters/ 2011MarketScanBibliography.pdf. Accessed May 9, 2011. SAS Institute Inc. The LOGISTIC procedure. In: SAS/STAT 9.2 User’s Guide. 2nd ed. Cary, North Carolina: SAS Institute Inc., 2009: 3255–3476. SAS Institute Inc. The PHREG procedure. In: SAS/STAT 9.2 User’s Guide. 2nd ed. Cary, North Carolina: SAS Institute Inc., 2009:4521– 4730. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, New York: John Wiley; 2000. SAS Institute Inc. The LIFETEST procedure. In: SAS/STAT 9.2 User’s Guide. 2nd ed. Cary, North Carolina: SAS Institute Inc., 2009: 3099 –3186. Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV, Singer DE. Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA) study. JAMA 2001;285:2370 –2375. Camm AJ, Kirchhof P, Lip GY, Schotten U, Savelieva I, Ernst S, van Gelder I, Al-Attar N, Hindricks G, Prendergast B, Heidbuchel H, Alfieri O, Angelini A, Atar D, Colonna P, De Caterina R, De Sutter J, Goette A, Gorenek B, Heldal M, Hohloser SH, Kolh P, Le Heuzey JY, Ponikowski P, Rutten FH, Vahanian A, Auricchio A, Bax J, Ceconi C, Dean V, Filippatos G, Funck-Brentano C, Hobbs R, Kearney P, McDonagh T, Popescu BA, Reiner Z, Sechtem U, Sirnes PA, Tendera M, Vardas PE, Widimsky P, Vardas PE, Agladze V, Aliot E, Balabanski T, Blomstrom-Lundqvist C, Capucci A, Crijns H, Dahlof B, Folliguet T, Glikson M, Goethals M, Gulba DC, Ho SY, Klautz RJ, Kose S, McMurray J, Perrone Filardi P, Raatikainen P, Salvador MJ, Schalij MJ, Shpektor A, Sousa J, Stepinska J, Uuetoa H, Zamorano JL, Zupan I. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Eur Heart J 2010;31:2369 –2429.
1533
19. Reynolds MR, Essebag V, Zimetbaum P, Cohen DJ. Healthcare resource utilization and costs associated with recurrent episodes of atrial fibrillation: the FRACTAL registry. J Cardiovasc Electrophysiol 2007;18:628 – 633. 20. Wolf PA, Mitchell JB, Baker CS, Kannel WB, D’Agostino RB. Impact of atrial fibrillation on mortality, stroke, and medical costs. Arch Intern Med 1998;158:229 –234. 21. Wu EQ, Birnbaum HG, Mareva M, Tuttle E, Castor AR, Jackman W, Ruskin J. Economic burden and co-morbidities of atrial fibrillation in a privately insured population. Curr Med Res Opin 2005;21:1693– 1699. 22. Hohnloser SH, Crijns HJ, van Eickels M, Gaudin C, Page RL, TorpPedersen C, Connolly SJ. Effect of dronedarone on cardiovascular events in atrial fibrillation. N Engl J Med 2009;360:668 – 678. 23. Wann LS, Curtis AB, January CT, Ellenbogen KA, Lowe JE, Estes NA III, Page RL, Ezekowitiz MD, Slotwiner DJ, Jackman WM, Stevenson WG, Tracy CM. 2011 ACCF/AHA/HRS focused update on the management of patients with atrial fibrillation (updating the 2006 guideline): a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation 2011;123:104 –123. 24. Kober L, Torp-Pedersen C, McMurray JJ, Gotzsche O, Levy S, Crijns H, Amlie J, Carlsen J. Increased mortality after dronedarone therapy for severe heart failure. N Engl J Med 2008;358:2678 –2687. 25. Sanofi provides Multaq phase IIIb PALLAS trial update. Available at: http://en.sanofi.com/binaries/20110707_PALLAS_en_tcm28 –33021. pdf. Accessed July 13, 2011. 26. Alonso A, Agarwal SK, Soliman EZ, Ambrose M, Chamberlain AM, Prineas RJ, Folsom AR. Incidence of atrial fibrillation in whites and African-Americans: the Atherosclerosis Risk in Communities (ARIC) study. Am Heart J 2009;158:111–117. 27. Austin PC, Daly PA, Tu JV. A multicenter study of the coding accuracy of hospital discharge administrative data for patients admitted to cardiac care units in Ontario. Am Heart J 2002;144:290 –296. 28. Borzecki AM, Bridgers DK, Liebschutz JM, Kader B, Kazis LE, Berlowitz DR. Racial differences in the prevalence of atrial fibrillation among males. J Natl Med Assoc 2008;100:237–245. 29. Boulanger L, Hauch O, Friedman M, Foster T, Dixon D, Wygant G, Menzin J. Warfarin exposure and the risk of thromboembolic and major bleeding events among Medicaid patients with atrial fibrillation. Ann Pharmacother 2006;40:1024 –1029.
a Penn State Heart & Vascular Institute, Penn State Milton S. Hershey Medical Center, Hershey, Pennsylvania; bQuintiles, Hawthorne, New York; and cSanofi-Aventis U.S., Bridgewater, New Jersey. Manuscript received November 18, 2011; revised manuscript received and accepted January 12, 2012. This study was funded by Sanofi-Aventis U.S., Bridgewater, New Jersey. Editorial support for this report was provided by Peloton Advantage, LLC, Parsippany, New Jersey, funded by Sanofi-Aventis U.S. *Corresponding author: Tel: 717-531-3907; fax: 717-531-4077. E-mail address: [email protected] (G.V. Naccarelli).
0002-9149/12/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.amjcard.2012.01.371
emic attack [doubled]) and CHA2DS2-VASc (congestive heart failure; hypertension; age ⱖ75 years [doubled]; type 2 diabetes; previous stroke, transient ischemic attack, or thromboembolism [doubled]; vascular disease; age 65 to 75 years; and sex category) scores.8 –11 The objective of this evaluation was to determine whether specific patient characteristics or risk factors, including CHADS2 and CHA2DS2-VASc scores, were predictive of first CV hospitalization in patients with AF or AFL. Methods This retrospective cohort evaluation used integrated patient data for the period from January 1, 2003, to June 30, 2009, obtained from the Thomson Reuters (Cambridge, Massachusetts) MarketScan Commercial Claims and Encounters database and Medicare Supplemental (January 1, 2003, to December 31, 2008) database (Figure 1). This database contains detailed enrollment, clinical utilization, and expenditure data for approximately 85 million subjects from approximately 45 large employers, health plans, and government and public organizations. These data reflect real-world treatment and outcomes and have been used for ⬎400 peer-reviewed publications.12 In accordance with the Health Insurance Portability and Accountability Act of 1996, all patient data were deidentified before analysis. The study population included patients aged ⱖ40 years www.ajconline.org
Review/CHADS2/CHA2DS2-VASc and Hospitalization in AF
1527
MarketScan Enrollees 01/01/2003–6/30/2009 N=84,744,071
Inclusion Criteria • AF index date between 01/01/2004–6/29/2009 (Medicaid: 12/30/2008) o Included: n=1,121,251 (1.3%); Excluded: n=83,652,820 (98.7%) • 12 months continuous enrollment prior to AF index date o Included: n=559,300 (49.9%); Excluded: n=561,951 (50.1%) • ≥1 day of information following AF index date and continuous enrollment during the entire follow-up period o Included: n=500,612 (89.5%); Excluded: n=58,688 (10.5%) • ≥40 years of age on AF index date o Included: n=492,446 (98.4%); Excluded: n=8,166 (1.6%) Exclusion Criteria • Inpatient or outpatient procedure codes related to cardiac surgery (excluding codes related to pacemaker and ablation) ≤30 days prior to AF index date o Patients meeting exclusion criteria: n=114,638 (23.3%) Study Population N=377,808 Inpatient/ER AF patients: n=199,412 (52.8%) Outpatient AF patients: n=178,396 (47.2%)
Yes
1st CV Hospitalization?
n=128,048 (33.9%)
No
n=249,760 (66.1%)
Figure 1. Patient identification. ER ⫽ emergency room.
with diagnoses of AF or AFL (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9CM] diagnosis codes 427.31 and 427.32) during ⱖ1 inpatient admission or ⱖ2 outpatient medical claims from January 1, 2004, to June 29, 2009 (December 30, 2008, for Medicaid patients). The qualifying diagnoses could appear in any position on a claim for any inpatient admission or within 30 days of each other for outpatient service. The date of the first qualifying AF diagnosis from January 1, 2004, to June 29, 2009, was designated the index date. Data from all identified patients were then further evaluated and included in the analysis if the patient was aged ⱖ40 years on the index date, had ⱖ12 months of continuous enrollment before the index date, had ⱖ1 day of follow-up information after the index date, had continuous enrollment during the entire follow-up period, and did not have an ICD-9-CM code relating to cardiac surgery (excluding pacemaker and ablation codes) 30 days before the index date. The study consisted of baseline and follow-up periods. Data from the baseline period, which started 364 days before the index date and ended on the index date, were used to obtain information about each patient’s medical history. Data from the follow-up period, which started the day after
the index date and ended after 24 months of observation, when the patient exited the database, or on the study end date (whichever came first), were used to assess hospitalization risk. The primary outcome of interest was the first CV hospitalization after the index date with all-cause CV hospitalization defined by ICD-9-CM codes 390.xx to 459.xx. A secondary outcome measure was elapsed time from index date to first CV hospitalization (time to first hospitalization). Demographic and clinical characteristics at baseline were summarized using descriptive statistics (means, medians, standard deviations, and ranges for continuous variables and frequencies and percentages for categorical variables). Univariate and multivariate analyses using logistic regression were performed to identify predictors of CV hospitalization, defined as binary outcomes (CV hospitalization, yes or no). Logistic regression modeling13 was used as an exploratory tool to better understand the interrelations among many potentially correlated clinical factors by estimating the odds ratio (OR) and 95% confidence interval (CI). Multivariate Cox proportional-hazard models14 were used to estimate hazard ratios (HR) and 95% CIs, to evaluate time to first CV hospitalization, and to understand the
1528
The American Journal of Cardiology (www.ajconline.org)
Table 1 Atrial fibrillation cohort demographic and baseline characteristics at index event date, MarketScan claims data, January 1, 2003, to June 30, 2009 Characteristic
Demographics Age at index (years) Median (range) Mean ⫾ SD Age category (years) 40–74 ⱖ75 Men Continuous enrollment (years) Median (range) Mean ⫾ SD AF diagnosis at index Inpatient Outpatient/ER Type of AF diagnosis New case Existing case Region Northeast North Central South West Unknown* Plan type Commercial Medicare Medicaid Baseline characteristics Charlson co-morbidity index 0 ⱖ1 Previous CV hospitalization Hypertension Stroke Myocardial infarction Pulmonary embolism Major bleeding Structural heart disease Other coronary artery disease Heart failure Cancer Valvular heart disease Cardiac hypertrophy Nonischemic cardiomyopathy Chronic obstructive pulmonary disease Type 2 diabetes Treated hyperthyroidism Composite AF symptom† CHADS2 score 0 1 2 3 4 5 6 CHA2DS2-VASc score 0 1 2
All Patients (n ⫽ 377,808)
76 (40–109) 73.9 ⫾ 12.1 171,026 (45.3%) 206,782 (54.7%) 196,840 (52.1%) 3.5 (1.01–6.5) 4.1 ⫾ 1.8
Patients With First CV Hospitalization (n ⫽ 128,048)
77 (40–108) 74.9 ⫾ 11.5 53,140 (41.5%) 74,908 (58.5%) 66,179 (51.7%) 4.0 (1.03–6.5) 4.3 ⫾ 1.7
Patients With No CV Hospitalizations (n ⫽ 249,760)
75 (40–109) 73.4 ⫾ 12.4 117,886 (47.2%) 131,874 (52.8%) 130,661 (52.3%) 3.5 (1.01–6.5) 4.0 ⫾ 1.9
199,412 (52.8%) 178,396 (47.2%)
72,076 (56.3%) 55,972 (43.7%)
127,336 (51.0%) 122,424 (49.0%)
238,443 (63.1%) 139,365 (36.9%)
59,004 (46.1%) 69,044 (53.9%)
179,439 (71.8%) 70,321 (28.2%)
31,578 (8.4%) 113,261 (30.0%) 106,797 (28.3%) 87,889 (23.3%) 38,283 (10.1%)
10,854 (8.5%) 40,294 (31.5%) 34,758 (27.1%) 29,786 (23.3%) 12,356 (9.6%)
20,724 (8.3%) 72,967 (29.2%) 72,039 (28.8%) 58,103 (23.3%) 25,927 (10.4%)
85,689 (22.7%) 254,664 (67.4%) 37,455 (9.9%)
23,488 (18.3%) 92,471 (72.2%) 12,089 (9.4%)
62,201 (24.9%) 162,193 (64.9%) 25,366 (10.2%)
203,089 (53.8%) 174,719 (46.2%) 242,740 (64.2%) 218,073 (57.7%) 54,504 (14.4%) 20,310 (5.4%) 7,616 (2.0%) 70,682 (18.7%) 191,402 (50.7%) 141,741 (37.5%) 117,647 (31.1%) 113,449 (30.0%) 99,140 (26.2%) 50,047 (13.2%) 28,844 (7.6%) 77,053 (20.4%) 76,735 (20.3%) 30,686 (8.1%) 120,964 (32.0%)
59,499 (46.5%) 68,549 (53.5%) 89,611 (70.0%) 77,778 (60.7%) 20,682 (16.2%) 7,830 (6.1%) 2,666 (2.1%) 26,784 (20.9%) 74,469 (58.2%) 56,168 (43.9%) 49,404 (38.6%) 39,241 (30.6%) 37,402 (29.2%) 20,628 (16.1%) 12,254 (9.6%) 31,448 (24.6%) 30,584 (23.9%) 10,933 (8.5%) 44,110 (34.4%)
143,590 (57.5%) 106,170 (42.5%) 153,129 (61.3%) 140,295 (56.2%) 33,822 (13.5%) 12,480 (5.0%) 4,950 (2.0%) 43,898 (17.6%) 116,933 (46.8%) 85,573 (34.3%) 68,243 (27.3%) 74,208 (29.7%) 61,738 (24.7%) 29,419 (11.8%) 16,590 (6.6%) 45,605 (18.3%) 46,151 (18.5%) 19,753 (7.9%) 76,854 (30.8%)
50,615 (13.4%) 105,004 (27.8%) 109,002 (28.9%) 66,172 (17.5%) 31,139 (8.2%) 13,053 (3.5%) 2,823 (0.7%)
11,598 (9.1%) 31,539 (24.6%) 39,235 (30.6%) 26,607 (20.8%) 12,365 (9.7%) 5,462 (4.3%) 1,242 (1.0%)
39,017 (15.6%) 73,465 (29.4%) 69,767 (27.9%) 39,565 (15.8%) 18,774 (7.5%) 7,591 (3.0%) 1,581 (0.6%)
15,776 (4.2%) 34,867 (9.2%) 54,564 (14.4%)
3,051 (2.4%) 8,063 (6.3%) 15,738 (12.3%)
12,725 (5.1%) 26,804 (10.7%) 38,826 (15.5%)
Review/CHADS2/CHA2DS2-VASc and Hospitalization in AF
1529
Table 1 (Continued) All Patients (n ⫽ 377,808)
Patients With First CV Hospitalization (n ⫽ 128,048)
Patients With No CV Hospitalizations (n ⫽ 249,760)
75,261 (19.9%) 80,105 (21.2%) 59,419 (15.7%) 35,069 (9.3%) 15,959 (4.2%) 5,681 (1.5%) 1,107 (0.3%)
24,911 (19.5%) 29,215 (22.8%) 23,343 (18.2%) 14,179 (11.1%) 6,632 (5.2%) 2,408 (1.9%) 508 (0.4%)
50,350 (20.2%) 50,890 (20.4%) 36,076 (14.4%) 20,890 (8.4%) 9,327 (3.7%) 3,273 (1.3%) 599 (0.2%)
Characteristic 3 4 5 6 7 8 9
* Includes Medicaid population of 37,455. Includes chest pain, tachycardia, palpitations, dizziness, syncope, dyspnea, and fatigue at AF index date.
†
Table 2 Atrial fibrillation cohort treatments at baseline, MarketScan claims data, January 1, 2003, to June 30, 2009 Treatment at Baseline
Rhythm-control drugs Rate-control drugs Antithrombotic drugs Statins Antihypertensive drugs Electrical cardioversion Ablation (CPT code 93651) Pacemaker Other cardiac surgery
All Patients (n ⫽ 377,808)
Patients With First CV Hospitalization (n ⫽ 128,048)
Patients With No CV Hospitalizations (n ⫽ 249,760)
16,443 (4.3%) 106,481 (28.2%) 88,444 (23.4%) 145,865 (38.6%) 197,263 (52.2%) 16,007 (4.2%) 5,422 (1.4%) 31,295 (8.2%) 175,243 (46.3%)
6,186 (4.8%) 40,838 (31.9%) 34,914 (27.3%) 52,280 (40.8%) 73,643 (57.5%) 5,840 (4.6%) 1,923 (1.5%) 12,644 (9.9%) 60,988 (47.6%)
10,257 (4.1%) 65,643 (26.3%) 53,530 (21.4%) 93,585 (37.5%) 123,620 (49.5%) 10,167 (4.1%) 3,499 (1.4%) 18,651 (7.5%) 114,255 (45.7%)
CPT ⫽ Current Procedural Terminology.
time dependency of various covariates (co-morbidities, treatments, demographics). All multivariate models used a backward selection method to remove statistically insignificant covariates, with a significance level ⱕ0.01 (the threshold required for any covariate to remain in the final model). Model fits were evaluated by Hosmer-Lemeshow test and C statistic.15 On the basis of clinical considerations, we reduced the number of initial variables (n ⫽ 36) to include the top 10 covariates. The life table method16 was used to estimate the percentage risk for hospitalization at different risk periods of 3, 6, 9, 12, 15, 18, 21, and 24 months for each respective CHADS2 and CHA2DS2-VASc score. For these models, risk was assessed at the end of each period. Results Of the nearly 85 million enrollees in the MarketScan database, 377,808 (0.45%) were identified who fulfilled all inclusion and exclusion criteria (Figure 1). Table 1 lists demographic and clinical characteristics at baseline for the overall study cohort and by CV hospitalization status. Differences between patients with first CV hospitalizations and no CV hospitalizations were statistically significant (p ⬍0.05) for all characteristics. Baseline CHADS2 and CHA2DS2-VASc factors were the strongest predictors of first CV hospitalization in patients with AF or AFL in all univariate and multivariate logistic and Cox proportional-hazards models. AF treatments and certain concomitant medications (listed in Table 2) were not
found to be predictive of first hospitalization. Therefore, models that included key co-morbidities provided the best clinical and statistical predictive results. On the basis of multivariate logistic models (age modeled as a continuous variable), congestive heart failure (OR 6.6, 95% CI 6.01 to 7.32), type 2 diabetes (OR 2.4, 95% CI 2.17 to 2.73), stroke (OR 3.1, 95% CI 2.68 to 3.57), and hypertension (OR 1.5, 95% CI 1.35 to 1.62) were found to increase the risk for first CV hospitalization (data not shown). In multivariate logistic regression models that included all variables (with age as a binary variable), the risk for first CV hospitalization was 3 times higher in patients with preexisting AF during baseline compared to patients with newly diagnosed AF. Therefore, estimates of time to first CV hospitalization were stratified by this variable. The individual components of CHADS2 (congestive heart failure: HR 1.7, 95% CI 1.65 to 1.71; type 2 diabetes: HR 1.3, 95% CI 1.28 to 1.33; stroke: HR 1.2, 95% CI 1.17 to 1.22; hypertension: HR 1.2, 95% CI 1.20 to 1.24) were also found to be associated with time to first CV hospitalization risk. However, because of violations in the underlying assumption that the HR is independent of time and valid only for time-independent covariates for the Cox proportional-hazards models, a nonparametric life table approach was taken to estimate the association of CHADS2 and CHA2DS2-VASc scores with first CV hospitalization risk. The correlation between CHADS2 and CHA2DS2-VASc
1530
The American Journal of Cardiology (www.ajconline.org)
higher) for scores ⬍2, but at higher scores, the differences in risk between these patients became more similar. In other words, the risk for CV hospitalization was similar between patients with new AF and those with preexisting AF if their baseline CHADS2 or CHA2DS2-VASc scores were already high. Discussion
Figure 2. Time to first CV hospitalization by CHADS2 (A) and CHA2DS2VASc (B) scores.
scores and risk for first CV hospitalization was present and relatively consistent from the first through the last month of the follow-up period (Figure 2). Relative to a CHADS2 score of 0, a score of 6 was associated with an excess 2.7-fold risk at 6 or 12 months, 2.5-fold risk at 18 months, and 2.3-fold risk at 24 months. A CHA2DS2-VASc score of 9 was associated with an approximate 3.0-fold increase in risk across time periods (6, 12, 18, and 24 months) compared to a score of 0 (Table 3). Patients who entered the cohort with preexisting AF had higher risks (OR 3.0, 95% CI 2.93 to 3.08) for CV hospitalization, even after adjustment for all other factors (treatment, age, other baseline co-morbidities listed in Table 1). For CHADS2 and CHA2DS2-VASc, the risk for CV hospitalization differed by whether patients were newly diagnosed or had preexisting diagnoses of AF at study entry (Figures 3 and 4). Relative to patients with preexisting diagnoses, the absolute risk for first CV hospitalization in newly diagnosed patients increased as CHADS2 and CHA2DS2-VASc scores increased, although this increase was constant over time. The differences in CV hospitalization risk between new and preexisting AF cases within a score category were greatest (relative risk about twofold
Approximately 4% of individuals aged ⱖ60 years in the United States have AF.17 This arrhythmia significantly increases the risk for stroke, other thromboembolic events, heart failure, hospitalization, and death.18 Approximately 31%, 48%, and 59% of patients with AF will be hospitalized for CV reasons (predominantly acute coronary syndromes, decompensated heart failure, thromboembolic events, and arrhythmia management) within 1, 3, and 5 years of diagnosis, respectively.4,18,19 These hospitalizations are the single largest contributor to the economic cost of AF.19 –21 Data indicate that the costs related to hospitalization in patients with AF equal or exceed those for all other medical care of these patients combined.7,19 –21 In a retrospective, claims-based analysis, annual direct medical costs for patients with and without AF were $15,553 and $2,792, respectively, with inpatient hospital costs ($8,486) accounting for 55% of the total direct cost of care in patients with AF.21 Similarly, in another retrospective, claims-based analysis, mean annual total health care costs for patients with and without AF were $23,750 and $7,439, respectively, with hospitalization costs ($13,507) accounting for 57% of the total health care costs in patients with AF.7 Risk factors for stroke, thromboembolism, and death have been extensively evaluated in patients with AF, and these evaluations have led to the development and validation of the CHADS2 and CHA2DS2-VASc scoring systems in an attempt to distinguish patients who might benefit from more intense antithrombotic therapy.8 –11 In contrast, only limited data are available to help identify patients with AF who are at high risk for hospitalization. A retrospective, community-based evaluation of approximately 4,500 individuals in Olmsted County, Minnesota, identified age, body mass index, systolic blood pressure, paroxysmal AF, and histories of myocardial infarction, valvular heart disease, peripheral artery disease, carotid artery disease, stroke, systemic hypertension, diabetes mellitus, chronic renal disease, and chronic obstructive pulmonary disease as independent risk factors for hospitalization in patients with AF.4 The present retrospective cohort evaluation is the largest and most geographically diverse assessment of hospitalization risk in patients with AF or AFL performed to date. In this assessment, CHADS2 and CHA2DS2-VASc scores were the 2 single best predictors of hospitalization risk in logistic regression and Cox proportional-hazards models. A CHADS2 score of 6 was associated with an approximate 2.5-fold increase, and a CHA2DS2-VASc score of 9 was associated with an approximate 3.0-fold increase in hospitalization risk relative to a score of 0, with comparatively proportional increments in risk for intermediate scores. Moreover, the predictive value of these scores was consistent throughout the evaluation period. Differences in CV
Review/CHADS2/CHA2DS2-VASc and Hospitalization in AF
1531
Table 3 Risk and 95% confidence intervals of first cardiovascular hospitalization by CHADS2 and CHA2DS2-VASc scores over time Score
CHADS2 0 1 2 3 4 5 6 CHA2DS2-VASc 0 1 2 3 4 5 6 7 8 9
% Risk (95% CI) 6 Months
12 Months
18 Months
24 Months
13.9 (13.58–14.21) 17.7 (17.47–17.96) 22.3 (22.05–22.59) 27.8 (27.42–28.18) 28.3 (27.73–28.85) 32.3 (31.42–33.26) 37.0 (34.92–39.13)
20.8 (20.42–21.20) 27.9 (27.63–28.24) 35.7 (35.34–36.00) 43.2 (42.75–43.66) 44.0 (43.34–44.68) 49.5 (48.43–50.59) 56.6 (54.25–59.06)
26.4 (25.98–26.88) 35.7 (35.35–36.05) 45.1 (44.77–45.51) 53.1 (52.62–53.60) 53.8 (53.03–54.49) 59.9 (58.80–61.10) 66.9 (64.42–69.39)
32.0 (31.47–32.55) 43.0 (42.58–43.40) 53.2 (52.80–53.65) 61.7 (61.11–62.22) 62.4 (61.52–63.18) 68.8 (67.50–70.14) 74.4 (71.51–77.12)
12.5 (11.95–13.03) 15.0 (14.63–15.42) 17.5 (17.19–17.87) 19.8 (19.49–20.10) 22.5 (22.22–22.84) 26.0 (25.61–26.39) 28.3 (27.75–28.81) 31.1 (30.26–31.90) 33.4 (32.00–34.83) 38.6 (35.33–42.08)
18.3 (17.62–18.95) 21.9 (21.37–22.34) 26.7 (26.24–27.07) 31.3 (30.88–31.64) 36.1 (35.69–36.46) 41.4 (40.90–41.84) 44.5 (43.89–45.16) 48.4 (47.47–49.40) 51.2 (49.61–52.91) 58.0 (54.26–61.80)
22.7 (21.98–23.53) 27.1 (26.60–27.71) 33.9 (33.45–34.40) 39.7 (39.30–40.14) 45.5 (45.08–45.94) 51.5 (51.03–52.06) 54.8 (54.10–55.48) 58.6 (57.52–59.59) 62.0 (60.25–63.74) 68.8 (64.93–72.67)
27.0 (26.13–27.99) 32.3 (31.62–32.93) 40.7 (40.18–41.30) 47.5 (47.02–48.00) 54.0 (53.46–54.45) 59.9 (59.27–60.44) 63.9 (63.12–64.70) 67.4 (66.21–68.55) 69.5 (67.48–71.42) 76.1 (71.54–80.37)
Figure 3. Time to first CV hospitalization by CHADS2 score stratified by preexisting (A) or new (B) AF diagnosis at index date.
Figure 4. Time to first CV hospitalization by CHA2DS2-VASc score stratified by preexisting (A) or new (B) AF diagnosis at index date.
1532
The American Journal of Cardiology (www.ajconline.org)
hospitalization risks with increasing scores was observed by type of patient diagnosis (those with preexisting AF diagnoses were at greater risk than those newly diagnosed with AF). Although the C statistics in these models (0.57 to 0.58) indicate only modest strength for predicting CV hospitalization risk, these statistics are similar to those observed for the more established use of these scoring systems in predicting thromboembolism risk.11 One possible explanation for the paucity of studies evaluating hospitalization risk in patients with AF is that until recently, there was little evidence that treatment could ameliorate this risk. The advantages of sinus rhythm in early clinical trials of AF therapy appear to have been offset by the deleterious effects of the antiarrhythmic drugs used, and hospitalization was more frequent in the rhythm-control arms of these trials.9 However, recent data from A PlaceboControlled, Double-Blind, Parallel Arm Trial to Assess the Efficacy of Dronedarone 400 mg b.i.d. for the Prevention of CV Hospitalization or Death from any Cause in Patients With Atrial Fibrillation/Atrial Flutter (ATHENA) demonstrated that treatment can safely and effectively reduce CV hospitalization risk.22 As a result, American and European treatment guidelines have recently been amended to recommend consideration of dronedarone use in patients with nonpermanent AF to reduce the risk for CV hospitalizations.18,23 The reduction in CV hospitalization with dronedarone appears to be limited to the ATHENA population, as patients with decompensated heart failure who met Antiarrhythmic Trial With Dronedarone in Moderate-to-Severe CHF Evaluating Morbidity Decrease (ANDROMEDA) inclusion criteria were excluded.22,24 Furthermore, the Permanent Atrial Fibrillation Outcome Study Using Dronedarone on Top of Standard Therapy (PALLAS) trial, designed to assess the efficacy of dronedarone compared to placebo in patients with permanent AF, was terminated prematurely because of a significant increase in CV events in the dronedarone arm.25 Further studies assessing the benefits of pharmacologic and nonpharmacologic rhythm control strategies in patients with AF are needed. Finally, it is important to consider limitations of the present evaluation. The evaluation was performed using information obtained from an administrative claims database. These data are typically collected to document justification for billing, not medical research, and thus clinical details, such as the type and severity of valvular disease, are often limited or unavailable. In addition, by relying on ICD-9-CM codes and National Drug Codes to define the study cohort, co-morbidities, drugs, and hospitalization, our analysis is vulnerable to inaccurate and/or differential coding practices of the sources providing information to the database. However, we used validated algorithms from the published research to minimize misclassification of AF; these algorithms have been found in multiple studies to have a high positive predictive value.26 –28 To further enhance specificity, we required ⱖ2 outpatient AF diagnoses occurring on different days or 1 inpatient diagnosis. Separating encounters by a minimal period (e.g., 30 days) has been used to increase specificity and identify chronic AF.29 Last, because of the large sample sizes in this analysis, small differences in demographics and baseline characteristics between patients with CV hospitalizations and those with no
CV hospitalizations resulted in significant p values that are clinically insignificant. Even covariates with ORs or HRs close to 1 had statistically significant p values because of the large sample size. Consequently, clinical and etiologic considerations guided selection of covariates and interpretation of modeling results. In summary, the risk for first CV hospitalization increased with time and CHADS2 and CHA2DS2-VASc scores in this retrospective cohort evaluation of patients with AF or AFL. As a result, this scoring system may be useful to clinicians in identifying patients at high risk for hospitalization who might benefit from targeted therapy to reduce this risk. Further investigation is warranted to validate these findings in other clinical and epidemiologic studies. Acknowledgment: We thank Gerald Barber, MD, for writing assistance. 1. Naccarelli GV, Varker H, Lin J, Schulman KL. Increasing prevalence of atrial fibrillation and flutter in the United States. Am J Cardiol 2009;104:1534 –1539. 2. Wattigney WA, Mensah GA, Croft JB. Increasing trends in hospitalization for atrial fibrillation in the United States 1985 through 1999: implications for primary prevention. Circulation 2003;108:711–716. 3. Miyasaka Y, Barnes ME, Gersh BJ, Cha SS, Bailey KR, Abhayaratna WP, Seward JB, Tsang TS. Secular trends in incidence of atrial fibrillation in Olmsted County, Minnesota 1980 to 2000, and implications on the projections for future prevalence. Circulation 2006;114: 119 –125. 4. Miyasaka Y, Barnes ME, Gersh BJ, Cha SS, Bailey KR, Seward JB, Tsang TS. Changing trends of hospital utilization in patients after their first episode of atrial fibrillation. Am J Cardiol 2008;102:568 –572. 5. Roger VL, Go AS, Lloyd-Jones DM, Adams RJ, Berry JD, Brown TM, Carnethon MR, Dai S, De Simone G, Ford ES, Fox CS, Fullerton HJ, Gillespie C, Greenlund KJ, Hailpern SM, Heit JA, Ho PM, Howard VJ, Kissela BM, Kittner SJ, Lackland DT, Lichtman JH, Lisabeth LD, Makuc DM, Marcus GM, Marelli A, Matchar DB, McDermott MM, Meigs JB, Moy CS, Mozaffarian D, Mussolino ME, Nichol G, Paynter NP, Rosamond WD, Sorlie PD, Stafford RS, Turan TN, Turner MB, Wong ND, Wylie-Rosett J. Heart disease and stroke statistics—2011 update: a report from the American Heart Association. Circulation 2011;123:459 – 463. 6. Naccarelli GV, Johnston SS, Lin J, Patel PP, Schulman KL. Cost burden of cardiovascular hospitalization and mortality in ATHENAlike patients with atrial fibrillation/atrial flutter in the United States. Clin Cardiol 2010;33:270 –279. 7. Lee WC, Lamas GA, Balu S, Spalding J, Wang Q, Pashos CL. Direct treatment cost of atrial fibrillation in the elderly American population: a Medicare perspective. J Med Econ 2008;11:281–298. 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. Fuster V, Ryden LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen KA, Halperin JL, Kay GN, Le Huezey JY, Lowe JE, Olsson SB, Prystowsky EN, Tamargo JL, Wann LS. 2011 ACCF/AHA/HRS focused updates incorporated into the ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines developed in partnership with the European Society of Cardiology and in collaboration with the European Heart Rhythm Association and the Heart Rhythm Society. J Am Coll Cardiol 2011;57:e101– e198. 10. Crandall MA, Horne BD, Day JD, Anderson JL, Muhlestein JB, Crandall BG, Weiss JP, Osborne JS, Lappe DL, Bunch TJ. Atrial fibrillation significantly increases total mortality and stroke risk be-
Review/CHADS2/CHA2DS2-VASc and Hospitalization in AF
11.
12.
13.
14.
15. 16.
17.
18.
yond that conveyed by the CHADS2 risk factors. Pacing Clin Electrophysiol 2009;32:981–986. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the euro heart survey on atrial fibrillation. Chest 2010;137:263–272. Thompson Reuters. MarketScan studies, abbreviated bibliography. Available at: http://img.en25.com/Web/ThomsonReuters/ 2011MarketScanBibliography.pdf. Accessed May 9, 2011. SAS Institute Inc. The LOGISTIC procedure. In: SAS/STAT 9.2 User’s Guide. 2nd ed. Cary, North Carolina: SAS Institute Inc., 2009: 3255–3476. SAS Institute Inc. The PHREG procedure. In: SAS/STAT 9.2 User’s Guide. 2nd ed. Cary, North Carolina: SAS Institute Inc., 2009:4521– 4730. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, New York: John Wiley; 2000. SAS Institute Inc. The LIFETEST procedure. In: SAS/STAT 9.2 User’s Guide. 2nd ed. Cary, North Carolina: SAS Institute Inc., 2009: 3099 –3186. Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV, Singer DE. Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA) study. JAMA 2001;285:2370 –2375. Camm AJ, Kirchhof P, Lip GY, Schotten U, Savelieva I, Ernst S, van Gelder I, Al-Attar N, Hindricks G, Prendergast B, Heidbuchel H, Alfieri O, Angelini A, Atar D, Colonna P, De Caterina R, De Sutter J, Goette A, Gorenek B, Heldal M, Hohloser SH, Kolh P, Le Heuzey JY, Ponikowski P, Rutten FH, Vahanian A, Auricchio A, Bax J, Ceconi C, Dean V, Filippatos G, Funck-Brentano C, Hobbs R, Kearney P, McDonagh T, Popescu BA, Reiner Z, Sechtem U, Sirnes PA, Tendera M, Vardas PE, Widimsky P, Vardas PE, Agladze V, Aliot E, Balabanski T, Blomstrom-Lundqvist C, Capucci A, Crijns H, Dahlof B, Folliguet T, Glikson M, Goethals M, Gulba DC, Ho SY, Klautz RJ, Kose S, McMurray J, Perrone Filardi P, Raatikainen P, Salvador MJ, Schalij MJ, Shpektor A, Sousa J, Stepinska J, Uuetoa H, Zamorano JL, Zupan I. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Eur Heart J 2010;31:2369 –2429.
1533
19. Reynolds MR, Essebag V, Zimetbaum P, Cohen DJ. Healthcare resource utilization and costs associated with recurrent episodes of atrial fibrillation: the FRACTAL registry. J Cardiovasc Electrophysiol 2007;18:628 – 633. 20. Wolf PA, Mitchell JB, Baker CS, Kannel WB, D’Agostino RB. Impact of atrial fibrillation on mortality, stroke, and medical costs. Arch Intern Med 1998;158:229 –234. 21. Wu EQ, Birnbaum HG, Mareva M, Tuttle E, Castor AR, Jackman W, Ruskin J. Economic burden and co-morbidities of atrial fibrillation in a privately insured population. Curr Med Res Opin 2005;21:1693– 1699. 22. Hohnloser SH, Crijns HJ, van Eickels M, Gaudin C, Page RL, TorpPedersen C, Connolly SJ. Effect of dronedarone on cardiovascular events in atrial fibrillation. N Engl J Med 2009;360:668 – 678. 23. Wann LS, Curtis AB, January CT, Ellenbogen KA, Lowe JE, Estes NA III, Page RL, Ezekowitiz MD, Slotwiner DJ, Jackman WM, Stevenson WG, Tracy CM. 2011 ACCF/AHA/HRS focused update on the management of patients with atrial fibrillation (updating the 2006 guideline): a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation 2011;123:104 –123. 24. Kober L, Torp-Pedersen C, McMurray JJ, Gotzsche O, Levy S, Crijns H, Amlie J, Carlsen J. Increased mortality after dronedarone therapy for severe heart failure. N Engl J Med 2008;358:2678 –2687. 25. Sanofi provides Multaq phase IIIb PALLAS trial update. Available at: http://en.sanofi.com/binaries/20110707_PALLAS_en_tcm28 –33021. pdf. Accessed July 13, 2011. 26. Alonso A, Agarwal SK, Soliman EZ, Ambrose M, Chamberlain AM, Prineas RJ, Folsom AR. Incidence of atrial fibrillation in whites and African-Americans: the Atherosclerosis Risk in Communities (ARIC) study. Am Heart J 2009;158:111–117. 27. Austin PC, Daly PA, Tu JV. A multicenter study of the coding accuracy of hospital discharge administrative data for patients admitted to cardiac care units in Ontario. Am Heart J 2002;144:290 –296. 28. Borzecki AM, Bridgers DK, Liebschutz JM, Kader B, Kazis LE, Berlowitz DR. Racial differences in the prevalence of atrial fibrillation among males. J Natl Med Assoc 2008;100:237–245. 29. Boulanger L, Hauch O, Friedman M, Foster T, Dixon D, Wygant G, Menzin J. Warfarin exposure and the risk of thromboembolic and major bleeding events among Medicaid patients with atrial fibrillation. Ann Pharmacother 2006;40:1024 –1029.