- Email: [email protected]
Renal Impairment and Ischemic Stroke Risk Assessment in Patients With Atrial Fibrillation
Journal of the American College of Cardiology © 2013 by the American College of Cardiology Foundation Published by Elsevier Inc.
Vol. 61, No. 20, 2013 ISSN 0735-1097/$36.00 http://dx.doi.org/10.1016/j.jacc.2013.02.035
Heart Rhythm Disorders
Renal Impairment and Ischemic Stroke Risk Assessment in Patients With Atrial Fibrillation The Loire Valley Atrial Fibrillation Project Amitava Banerjee, MPH, DPHIL,* Laurent Fauchier, MD, PHD,† Patrick Vourc’h, MD,‡ Christian R. Andres, MD, PHD,‡ Sophie Taillandier, MD,† Jean Michel Halimi, MD, PHD,§ Gregory Y. H. Lip, MD* Birmingham, United Kingdom; and Tours, France Objectives
This study sought to determine the risk of ischemic stroke (IS)/thromboembolism (TE) associated with renal impairment and its incremental predictive value over established risk stratification scores (congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke [CHADS2] and congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke, vascular disease, age 65 to 74 years, sex category (female) [CHA2DS2-VASc]) in patients with atrial fibrillation (AF).
Background
Risk stratification schemes for prediction of IS/TE in patients with AF are validated but do not include renal impairment.
Methods
Patients diagnosed with nonvalvular AF and available estimated glomerular filtration rate (eGFR) data in a 4-hospital institution between 2000 and 2010 were identified. The study population was stratified by renal impairment defined by serum creatinine level and by eGFR measured at time of diagnosis of AF. Independent risk factors of IS/TE (including renal impairment) were investigated in Cox regression models. The incremental predictive value of renal impairment over CHADS2 and CHA2DS2-VASc were assessed with the c-statistic, net reclassification improvement, and integrated discrimination improvement. We focused on the 1-year outcomes in our analyses.
Results
Of 8,962 eligible individuals, 5,912 (66%) had nonvalvular AF and available eGFR data. Renal impairment by both creatinine and eGFR definitions was associated with higher rates of IS/TE at 1 year, compared with normal renal function. After adjustment for CHADS2 risk factors, renal impairment did not significantly increase the risk of IS/TE at 1 year (hazard ratio: 1.06; 95% confidence interval [CI]: 0.75 to 1.49 for renal impairment; and hazard ratio: 1.09; 95% CI: 0.84 to 1.41 for eGFR). When renal impairment was added to existing risk scoring systems for stroke/TE (CHADS2 and CHA2DS2-VASc), it did not independently add to the predictive value of the scores, whether defined by serum creatinine level or eGFR. This was evident even when the analysis was confined to only those patients with at least 1 year of follow-up.
Conclusions
Renal impairment was not an independent predictor of IS/TE in patients with AF and did not significantly improve the predictive ability of the CHADS2 or CHA2DS2-VASc scores. (J Am Coll Cardiol 2013;61:2079–87) © 2013 by the American College of Cardiology Foundation
Both atrial fibrillation (AF) and chronic kidney disease (CKD) are increasingly recognized as significant burdens of morbidity and mortality of global proportions (1–11). Even mild renal impairment has long-term consequences for
cardiovascular health outcomes (5–7). Individuals with CKD are more likely to develop AF (12,13) and ischemic stroke (IS)/thromboembolism (TE) (14) than patients with normal renal function. In a recent Danish prospective study
From the *University of Birmingham Centre for Cardiovascular Sciences, City Hospital, Birmingham, United Kingdom; †Service de Cardiologie, Pôle Coeur Thorax Vasculaire, Centre Hospitalier Regional Universitaire et Faculté de Médecine, Université François Rabelais, Tours, France; ‡Service de Biochimie, Centre Hospitalier Regional Universitaire et Faculté de Médecine, Université François Rabelais, Tours, France; and the §Service de Nephrologie-Immunologie Clinique, Centre Hospitalier Regional Universitaire et Faculté de Médecine, Université François Rabelais, Tours, France. Dr. Fauchier has served as a consultant for Bayer, Medtronic, and Sanofi Aventis; and has received funding for conference travel and educational
symposia from Boehringher Ingelheim, Bayer, Medtronic, and Sanofi Aventis. Dr. Lip has served as a consultant for Bayer, Astellas, Merck, AstraZeneca, Sanofi Aventis, Biotronik, BMS/Pfizer, and Boehringher Ingelheim; and has been on the Speakers’ Bureau for Bayer, BMS/Pfizer, Boehringher Ingelheim, and Sanofi Aventis. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Halimi and Lip contributed equally to this work. Manuscript received December 11, 2012; revised manuscript received February 11, 2013, accepted February 19, 2013.
2080
Banerjee et al. Renal Impairment and Stroke Risk in AF
of 132,372 individuals with AF, 3,587 individuals had CKD, which was associated with increased risk of AF ⴝ atrial fibrillation IS/TE and bleeding (15), confirmCHADS2 ⴝ congestive ing observations of previous smaller heart failure, hypertension, studies (14,16,17). This study also age >75 years, diabetes, previous stroke showed the benefit of vitamin K antagonist (VKA) therapy on CHA2DS2-VASc ⴝ congestive heart failure, IS/TE outcomes in the setting of hypertension, age >75 CKD, although both VKA and years, diabetes, previous aspirin were associated with an stroke, vascular disease, increased risk of bleeding (15). age 65 to 74 years, sex category (female) Renal function is quantified by CI ⴝ confidence interval urinary creatinine clearance or by the estimated glomerular filtraCKD ⴝ chronic kidney disease tion rate (eGFR) (18 –20), but eGFR ⴝ estimated few studies have considered the glomerular filtration rate association between eGFR and HAS-BLED ⴝ hypertension, long-term outcomes in individuabnormal renal/liver als with AF, and existing studies function, stroke, bleeding have tended to consider renal history or predisposition, function as a dichotomous varilabile international normalized ratio, elderly able (2– 4,14). Also, patients (>65 years), drugs/alcohol with renal failure have been exconcomitantly cluded from randomized trials of HR ⴝ hazard ratio IS prevention in AF. IDI ⴝ integrated The importance of risk predicdiscrimination improvement tion tools for IS/TE (congestive IS ⴝ ischemic stroke heart failure, hypertension, age NRI ⴝ net reclassification ⱖ75 years, diabetes, previous improvement stroke [CHADS2] [21]; and OAC ⴝ oral anticoagulation congestive heart failure, hyperTE ⴝ thromboembolism tension, age ⱖ75 years, diabetes, VKA ⴝ vitamin K previous stroke, vascular disease, antagonist age 65 to 74 years, sex category (females) [CHA 2 DS 2 -VASc] [22]) in prognosis and treatment planning in AF patients is recognized by their inclusion in major international guidelines (23,24). Renal failure is included as a dichotomous variable in risk prediction tools for bleeding but not included in risk prediction tools for IS/TE (25–27), although the possibility of adding renal impairment (as the little “c”) to the CHA2DS2-VASc score has been previously proposed (28,29). In a small study of selected AF patients post-catheter ablation, renal dysfunction— defined by eGFR ⬍60 ml/min/1.73 m2— independently increased the risk of TE (hazard ratio [HR]: 6.8; 95% confidence interval [CI]: 4.2 to 12.1) (30). Therefore, better understanding of the impact of renal function on IS/TE outcomes in a more representative “real-world” population of AF patients is required. The present study represents the first analysis to consider the association between renal function, as measured by serum creatinine level or eGFR, and the risk of IS/TE events in a “real-world” population of individuals with AF, unrestricted by age or comorbidity. We also investigated the incremental predictive value of adding renal Abbreviations and Acronyms
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
function to established IS risk scores in AF (CHADS2 and CHA2DS2-VASc).
Methods Study population. The methods of the Loire Valley Atrial Fibrillation Project have been previously reported (31). Patients were followed from the first record of AF after January 1, 2000 (i.e., index date) up to the latest data collection at the time of study (December 2010) (see the Methods section in the Online Appendix). Treatment at discharge was obtained by screening hospital stay reports, and information on comorbidities was obtained from the computerized coding system. Patients were excluded from the study if there were no available data with regard to the baseline serum creatinine level at the time of first diagnosis of AF (Fig. 1). For each patient, the CHADS2 (21); CHA2DS2-VASc (22); and hypertension, abnormal renal/ liver function, stroke, bleeding history or predisposition, labile international normalized ratio, elderly (⬎65 years), drugs/alcohol concomitantly (HAS-BLED) (25) scores were calculated (Table 1). Assessment of renal function. Serum creatinine levels were taken at baseline, which was at first diagnosis of AF. Index date was the time of the first record of AF and thus the serum creatinine (and hence eGFR) measurement was from that index date. Renal impairment was defined as reported history of renal failure or baseline serum creatinine level of ⬎133 mol/l in men and ⬎115 mol/l in women (32). To convert serum creatinine from mol/l to mg/dl, the former was multiplied by a conversion factor of 88.4. Current consensus guidelines state that prediction equations have greater consistency and accuracy than serum creatinine in the assessment of GFR (18 –20,33–35). In addition, prediction equations are equivalent or better than 24-h urine creatinine clearance in all but 1 study (19,20,36). The eGFR (ml/min/1.73 m2) was calculated (see the Methods section in the Online Appendix). Outcomes. During follow-up, information on outcomes of TE (including peripheral artery embolism and transient ischemic attack), stroke (ischemic or hemorrhagic), bleeding, and all-cause mortality were recorded. In this study, the outcome of interest was IS/TE. Hemorrhagic strokes were excluded in our analyses. Statistical analysis. The study population was stratified into 3 categories according to eGFR (in ml/min/1.73 m2), corresponding to the stages of CKD: ⱖ60, 30 to 59, and ⬍30 (Fig. 1) (18 –20). Because data with regard to proteinuria were not available, stage of renal impairment could not be defined. Baseline characteristics were determined separately for the 3 eGFR strata, and differences were investigated with the chi-square test for categorical covariates and the KruskalWallis test for continuous covariates. Baseline characteristics were also determined by the presence or absence of renal impairment and then further stratification by eGFR.
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
Figure 1
2081
Study Population
Patient characteristics shown, by renal impairment and by estimated glomerular filtration rate (eGFR). VKA ⫽ vitamin K antagonist.
First, overall rates of IS/TE were calculated for patients with CHADS2 and CHA2DS2-VASc scores. Thereafter, rates were determined separately for patients with and without renal impairment and by eGFR separately and then by renal impairment with further stratification by eGFR. Second, Cox proportional-hazard regression models were constructed to investigate whether renal impairment and eGFR were independent predictors of IS/TE. The risks associated with renal impairment and eGFR were estimated in a univariate analysis, a sex- and age-adjusted analysis, an analysis adjusted for the risk factors included in the CHADS2 score, and a multivariate analysis adjusted for all baseline characteristics in Table 1, which were statistically significant (p ⬍ 0.05). All analyses were repeated by eGFR category and by combined stratification by renal impairment and eGFR. Furthermore, to test whether the results were influenced by patients initiating treatment with VKA, we performed additional analyses excluding patients at the initiation of such treatment. The value of adding renal impairment and eGFR to the CHADS2 score was evaluated by Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI), as previously described (37). Finally, we also assessed the predictive capability of CHADS2 with c-statistics estimated from Cox regression models with the method described by Liu et al. (38). In estimating the c-statistics, the CHADS2 score was analyzed as categoric risk groups (low/intermediate/high), and the effect of adding renal impairment to the score was determined.
The NRI, IDI, and the c-statistic were used to test the impact of adding renal impairment to the 2 scoring systems (CHADS2 and CHA2DS2-VASc) in 4 different statistical models by adding: 1) 1 point for renal impairment; 2) 1 point for eGFR ⫽ 30 to 59 and 2 points for eGFR ⬍30; 3) 1 point for renal impairment and 2 points for renal impairment and eGFR ⬍30; and 4) 1 point for eGFR ⬍30. These different models use the most common definitions of renal function (by eGFR and serum creatinine) in different ways to add to CHADS2 and CHA2DS2-VASc in the most user-friendly way. This approach assesses which (if any) method of adding renal impairment to either of the stroke risk scores added predictive value. We felt it was important to explore the impact of renal (dys)function in different statistical models, because renal impairment is a continuum and not just a “yes/no” phenomenon. In short, the different tested models aimed to find the most user-friendly and useful stroke risk prediction tool incorporating renal function. A 2-sided p value ⬍0.05 was considered statistically significant. All analyses were performed with SPSS statistical software (version 18.0, IBM, Chicago, Illinois), and NRI and IDI calculations were performed with R statistical software (version 2.12.2, University of Auckland, Auckland, New Zealand). Results Of 8,962 eligible individuals, 5,912 (66%) had nonvalvular AF and available serum creatinine data, allowing the eGFR
2082
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
Characteristics of Patients With AF by Renal Impairment eGFR and eGFR Table 1 Characteristics of Patients With AF by Renal and Impairment Patients With Renal Impairment eGFR (ml/min/1.73 m2)
Overall Study Population Normal Renal Function (n ⴝ 4,375)
Renal Impairment (n ⴝ 1,537)
p Value
30–59 (n ⴝ 1,196)
<30 (n ⴝ 341)
Age, yrs
68.2 ⫾ 15.7
75.8 ⫾ 10.9
⬍0.001
74.9 ⫾ 10.8
79.3 ⫾ 9.5
Female
1,777 (40.6)
419 (27.3)
⬍0.001
234 (19.6)
185 (54.3)
⬍0.001
Paroxysmal
2,635 (60.2)
760 (49.4)
⬍0.001
583 (48.7)
177 (51.9)
0.49
Permanent
1501 (34.3)
688 (44.8)
541 (45.2)
147 (43.1)
Persistent
239 (5.5)
89 (5.8)
72 (6.0)
17 (5.0)
p Value
Type of AF
Comorbidities 1,780 (40.7)
757 (49.3)
⬍0.001
548 (45.8)
209 (61.3)
Diabetes
653 (14.9)
315 (20.5)
⬍0.001
220 (18.4)
95 (27.9)
0.001
Previous stroke
315 (7.2)
133 (8.7)
0.06
94 (7.9)
39 (11.4)
0.11
Coronary artery disease
1,215 (27.8)
631 (41.1)
⬍0.001
503 (42.1)
128 (37.5)
0.16
Any vascular disease
1,336 (30.5)
698 (45.4)
⬍0.001
549 (45.9)
149 (43.7)
0.33
Heart failure
2,083 (47.6)
1,015 (66.0)
⬍0.001
760 (63.5)
255 (74.8)
⬍0.001
Hypertension
Liver impairment Dyslipidemia
⬍0.001
14 (0.3)
4 (0.3)
0.72
3 (0.3)
1 (0.3)
0.99
821 (18.8)
327 (21.3)
0.03
254 (21.2)
73 (21.4)
0.76
Smoking
531 (12.1)
213 (13.9)
0.08
165 (13.8)
48 (14.1)
0.85
Pacemaker/ICD
689 (15.7)
349 (22.7)
⬍0.001
277 (23.2)
72 (21.1)
0.54
Bleeding risk factors 192 (4.4)
91 (5.9)
0.02
64 (5.4)
27 (7.9)
0.20
Labile INR
Previous bleeding
80 (1.8)
24 (1.6)
0.49
15 (1.3)
9 (2.6)
0.19
Anemia
21 (0.5)
16 (1.0)
0.02
11 (0.9)
5 (1.5)
0.68
NSAIDs
6 (0.1)
2 (0.1)
0.95
2 (0.2)
0 (0.0)
0.75
Drugs
717 (16.4)
365 (23.7)
⬍0.001
290 (24.2)
75 (22.0)
0.50
Cancer
72 (1.6)
32 (2.1)
0.26
22 (1.8)
10 (2.9)
0.45
Excessive risk of falls
40 (0.9)
23 (1.5)
0.06
15 (1.3)
8 (2.3)
0.34
4 (0.1)
1 (0.1)
0.76
1 (0.1)
0 (0.0)
0.87 ⬍0.001
Thrombocytopenia Antithrombotic agents Vitamin K antagonist
2,295 (52.5)
796 (51.8)
0.58
654 (54.7)
142 (46.9)
Antiplatelet
1,349 (30.8)
525 (34.2)
0.002
408 (34.1)
117 (39.4)
0.52
Any antithrombotic
3,238 (74.0)
1127 (73.3)
0.21
892 (74.6)
235 (78.1)
0.001
Heart failure therapy ACEI
779 (17.8)
350 (22.8)
0.06
283 (23.7)
67 (19.6)
0.02
Beta-blocker
981 (22.4)
425 (27.7)
0.16
344 (28.8)
81 (23.8)
0.004
Digoxin
563 (12.9)
206 (13.4)
0.11
166 (13.9)
40 (11.7)
0.14
Diuretic
807 (18.4)
467 (30.4)
⬍0.001
348 (29.1)
119 (34.9)
0.07
1,222 (27.9)
443 (28.8)
0.52
347 (29.0)
96 (28.2)
0.49
157 (3.6)
70 (4.6)
0.01
52 (4.3)
18 (5.3)
0.12
⬍0.001
⬍0.001
Antiarrhythmic agent Calcium channel blocker CHADS2 Low (score ⫽ 0)
1,084 (24.8)
127 (8.3)
122 (10.2)
5 (1.5)
Intermediate (score ⫽ 1)
1,119 (25.6)
373 (24.3)
318 (26.6)
55 (16.1)
High (score ⱖ2)
2,172 (49.6)
1037 (67.5)
756 (63.2)
281 (82.4)
CHA2DS2-VASc Low (score ⫽ 0)
475 (10.9)
38 (2.5)
37 (3.1)
1 (0.3)
Intermediate (score ⫽ 1)
682 (15.6)
91 (5.9)
86 (7.2)
5 (1.5)
3,218 (73.6)
1408 (91.6)
1073 (89.7)
335 (98.2)
Low (score ⫽ 0)
1,030 (23.5)
101 (6.6)
95 (7.9)
6 (1.8)
Moderate (score ⫽ 1–2)
1,628 (37.2)
544 (35.4)
480 (40.1)
64 (18.8)
High (score ⱖ3)
1,717 (39.2)
892 (58.0)
621 (51.9)
136 (39.9)
High (score ⱖ2)
⬍0.001
⬍0.001
HAS-BLED ⬍0.001
⬍0.001
Values are mean ⫾ SD or n (%). According to the congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke (CHADS2) and congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke, vascular disease, age 65 to 74 years, sex category (female) (CHA2DS2-VASc) scores, patients with a score of 0 on either schema were considered as “low risk,” 1 as “intermediate risk,” and ⱖ2 as “high risk” of stroke and thromboembolism. The Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile International Normalized Ratio, Elderly (⬎65 years), Drugs/alcohol concomitantly (HAS-BLED) score is a validated scoring system for bleeding risk stratification in atrial fibrillation (AF) patients (25). Patients with HAS-BLED score of 0 to 2 were deemed to have “low-moderate” bleeding risk, and those with HAS-BLED score of ⱖ3 were classified as “high” bleeding risk. CHADS2 (1 point each for heart failure, hypertension, age ⱖ75 and diabetes, and 2 points for prior stroke or thromboembolism); CHA2DS2-VASc (1 point for heart failure, hypertension, diabetes, vascular disease, age 65 to 74, and female sex; 2 points for prior stroke or thromboembolism and age ⱖ75). ACEI ⫽ angiotensin-converting enzyme inhibitor; eGFR ⫽ estimated glomerular filtration rate; ICD ⫽ implantable cardiac defibrillator; INR ⫽ international normalized ratio.
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
Patients Rate of Stroke/TE and in Patients per 100 With Person-Years CHADS in Allⴝ 0 With and Without Renal Impairment Rate of Stroke/TE per 100 Person-Years in All 2 Score Table 2 Patients and in Patients With CHADS2 Score ⴝ 0 With and Without Renal Impairment 1-Yr Follow-Up
Person-Yrs
Stroke/TE Events
Stroke/TE Rate (CI)
Overall (whole cohort) Total
3,669
171
4.7 (3.5–6.2)
eGFR ⱖ60
1,863
64
3.4 (2.4–4.8)
eGFR ⫽ 30–59
1,610
92
5.7 (4.2–7.8)
196
15
7.7 (4.3–13.6)
2,738
119
4.4 (3.2–5.9)
Renal impairment
931
52
5.6 (3.9–8.1)
eGFR ⫽ 30–59
735
37
5.0 (3.3–7.6)
eGFR ⬍30
196
15
7.7 (4.3–13.6)
eGFR ⬍30 Normal renal function
CHADS2 ⫽ 0 Total
688
11
1.6 (0.8–3.0)
eGFR ⱖ60
521
9
1.7 (0.9–3.4)
eGFR ⫽ 30–59
163
2
1.2 (0.3–5.0)
4
0
—
614
10
1.6 (0.8–3.2)
Renal impairment
74
1
1.4 (0.2–9.8)
eGFR ⫽ 30–59
70
1
1.4 (0.2–10.4)
4
0
—
eGFR ⬍30 Normal renal function
eGFR ⬍30
CI ⫽ confidence interval; TE ⫽ thromboembolism; other abbreviations as in Table 1.
to be calculated (Fig. 1). Of the cohort, 3,944 individuals had ⱕ1 year of follow-up, whereas 1,444 individuals had 2 to 5 years of follow-up, and 524 had 6 to 10 years of follow-up. Thus, 14,499 patient-years of follow-up were included in the analysis, with mean follow-up of 2.45 ⫾ 3.56 years. We focused on the 1-year outcomes in our analyses. Of the cohort, 1,537 (26%) individuals had renal impairment, and of these, 341 (5.8%) had eGFR ⬍30 ml/min/ 1.73 m2 (Fig. 1). Online Figure 1 illustrates the study population by eGFR alone. Baseline characteristics are shown in Table 1 by renal impairment and eGFR category and in Online Table 1 by eGFR category. Individuals with renal impairment were older, more likely to be male, less likely to have paroxysmal AF, more likely to have comorbidities, and at higher risk of stroke/TE (assessed by CHADS2 and CHA2DS2-VASc scores) and bleeding (assessed by HAS-BLED score) (Table 1). Similar trends were present when those with renal impairment were further stratified by eGFR (Table 1) and when the overall population was stratified by eGFR (Online Table 1). Of 434 IS/TE events, 171 (39.4%) occurred within the first year. Table 2 shows the rates of stroke/TE per 100 person-years according to the presence or absence of renal impairment in patients with CHADS2 score ⫽ 0 and in the overall population. Subjects with normal renal function had lower event rates for IS/TE at 1 year. Patients with renal impairment had a higher rate of stroke/TE, and patients with eGFR ⬍30 ml/min/1.73 m2 had a higher rate than those patients with eGFR 30 to 59 ml/min/1.73 m2 in the overall population. Due to the small number of individuals
2083
with CHADS2 score ⫽ 0, there were no events in the eGFR ⬍30 ml/min/1.73 m2 category, and the rate of IS/TE could not be calculated for this particular category, and CIs for event rates in other subgroups of renal function were wide. Rates of stroke/TE were 3.3% and 7% at 1 year in individuals with eGFR ⱖ60 ml/min/1.73 m2 and with eGFR ⬍30 ml/min/1.73 m2, respectively. The corresponding rates of all-cause mortality at 1 year were 4.2% and 17.6%, respectively (Online Fig. 2). Online Table 2 displays the rates in patients with a CHA2DS2-VASc score ⫽ 0 to 1. Only 91 patients with renal impairment had a CHA2DS2-VASc score ⫽ 1 and only 2 events occurred in this subgroup. No statistically significant differences in IS/TE rates were observed between individuals when this subpopulation was stratified by renal function. Table 3 shows the results from the Cox regression analyses, showing the association of risk factors and risk of IS/TE. As a categoric variable (but not as a continuous variable), eGFR was an independent predictor of IS/TE in AF at 1 year follow-up—after adjustment for age, sex, and CHADS2 risk factors— but not for baseline characteristics. Renal impairment did not significantly increase the risk of IS/TE in univariate analyses or after adjustment for CHADS2 risk factors, age, sex, or baseline characteristics at 1 year. When considered as a continuous variable, eGFR was also not associated with an increased risk of IS/TE at 1 year. Online Table 3 shows analogous results from Cox regression analyses after excluding patients on a regimen of vitamin K antagonists at baseline (n ⫽ 3,592; 60.8%). Renal Results Renal Impairment FromRenal Cox and Regression Risk of Analyses Stroke/TE: Impairment and Risk of Stroke/TE: Table 3 Results From Cox Regression Analyses 1-Yr Follow-Up HR (95% CI) Univariate analysis Renal impairment
1.05 (0.76–1.46)
eGFR (categoric variable)
1.15 (0.90–1.47)
Adjusted for sex and age Renal impairment
1.03 (0.73–1.43)
eGFR (categoric variable)
1.07 (0.82–1.40)
Female
1.16 (0.84–1.61)
Age/5-yr increase
1.04 (0.99–1.02)
Adjusted for CHADS2 risk factors Renal impairment
1.06 (0.75–1.49)
eGFR (categoric variable)
1.09 (0.84–1.41)
Heart failure
0.91 (0.65–1.27)
Hypertension
1.22 (0.88–1.68)
Age ⱖ 75 yrs
1.25 (0.77–2.02)
Diabetes mellitus
1.22 (0.82–1.80)
Previous stroke/TE
0.88 (0.62–1.26)
Adjusted for baseline characteristics* Renal impairment
1.09 (0.77–1.55)
eGFR (categoric variable)
1.08 (0.83–1.41)
*Adjusted for Table 1 risk factors that were statistically significant (p ⬍ 0.05), and including age as a continuous covariate, the result is only displayed for renal impairment. HR ⫽ hazard ratio; other abbreviations as in Tables 1 and 2.
2084
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
In the overall study population, the c-statistic for the CHADS2 (HR: 0.64; 95% CI: 0.61 to 0.67) and CHA2DS2-VASC (HR: 0.64; 95% CI: 0.62 to 0.67) were similar. There was no statistically significant improvement in either the CHADS2 or the CHA2DS2-VASc scoring systems by the addition of renal function, regardless of the 4 models used to add renal function to the risk scores, or method of estimating reclassification (i.e., NRI or IDI) (Table 4). This was still evident even when the analysis was confined to only those patients with at least 1 year of follow-up (Online Table 4). Discussion
Figure 2
Risk of Stroke/Thromboembolism With Renal Impairment
Results from Cox regression analyses. CHADS2 ⫽ congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke; eGFR ⫽ estimated glomerular filtration rate.
impairment was not an independent predictor for IS/TE on univariate analysis (HR: 1.51; 95% CI: 0.93 to 2.44), after adjusting for age/sex (HR: 1.16; 95% CI: 0.71 to 1.90), CHADS2 risk factors (HR: 1.20; 95% CI: 0.73 to 1.96), or baseline characteristics (HR: 0.87; 95% CI: 0.46 to 1.65) at 1 year of follow-up. The Forest plots in Figure 2 illustrate the results from Cox regression analyses evaluating differences in IS/TE risk according to renal function, confirming that renal impairment and eGFR did not significantly increase the risk of IS/TE after univariate or multivariate models at 1 year. Table 4 illustrates the impact of adding renal impairment to the CHADS2 and CHA2DS2-VASc scores, as measured by NRI, IDI, and c-statistic. Online Table 4 shows the same analysis confined to only those patients with at least 1 year of follow-up (294 IS/TE events in 2,663 individuals over the study period).
In this study amongst a large “real-world” cohort of individuals with nonvalvular AF, we have shown for the first time that adding renal impairment to existing stroke risk stratification systems for IS/TE (CHADS2 and CHA2DS2VASc) did not independently add to the predictive value of these scores at 1 year of follow-up, whether it was defined by serum creatinine level or the eGFR or whether the analysis was confined to only those patients with at least 1 year of follow-up. Second, we show that renal impairment by both creatinine and eGFR definitions was associated with higher crude rates of stroke/TE at 1 year compared with those with normal renal function, but renal impairment was not independently associated with increased risk of stroke/TE at 1 year. Although it has been well-recognized that renal impairment is a risk factor for IS/TE in AF patients, renal impairment has not been included in the CHADS2 or CHA2DS2-VASc scores, due to lack of large prospective cohort data validating the additive value of renal impairment to these scores. Trial datasets would not answer this question, because severe renal impairment was often an exclusion criteria for clinical trials of IS prevention in AF. One highly selected cohort of AF patients undergoing ablation did suggest an additive value of renal dysfunction to
NRI and4IDI by Renal Impairment the CHADS andCHADS CHA2DS Scores 2VASc Table NRIAdding and IDI by Adding Renal to Impairment to2 the CHA2DS 2 and 2VASc Scores Patients Reclassified (%)
NRI
IDI
Patients With Event (n ⴝ 434)
Patients Without Event (n ⴝ 5,478)
NRI
SE
Z
p Value
IDI
AUC*
p Value
CHADS2 Model 1
31 (7.1)
469 (8.6)
⫺0.015
0.011
⫺0.03
0.49
0.0001
⫺0.004
0.41
CHADS2 Model 2
54 (12.4)
957 (17.5)
⫺0.051
0.010
0.27
0.61
⫺0.0001
⫺0.001
0.57
CHADS2 Model 3
31 (7.1)
469 (8.6)
⫺0.016
0.011
⫺0.08
0.47
⫺0.0001
⫺0.007
0.40
CHADS2 Model 4
5 (1.2)
55 (1.0)
0.002
0.011
0.54
0.71
0.0001
⬍0.0001
0.44
CHA2DS2VASc Model 1
2 (0.5)
127 (2.3)
⫺0.018
0.008
⫺0.90
0.18
⫺0.0001
⫺0.005
0.44
CHA2DS2VASc Model 2
2 (0.5)
238 (4.3)
⫺0.039
0.008
⫺0.80
0.21
0.0002
⫺0.014
0.30
CHA2DS2VASc Model 3
2 (0.5)
127 (2.3)
⫺0.018
0.008
⫺0.90
0.18
⫺0.0001
⫺0.005
0.44
CHA2DS2VASc Model 4
0 (0.0)
5 (0.1)
⫺0.001
0.009
⫺0.05
0.48
0.0000
⬍0.0001
0.25
The net reclassification improvement (NRI) focuses on reclassification tables constructed separately for patients with and without event and quantifies the correct movement in CHADS2 score by adding renal impairment (i.e., increased CHADS2 score for events by adding renal impairment counts positively, and increased CHADS2 score for non-events counts negatively). *Improvement in the area under the receiver operator characteristic curve (AUC) (c-statistic) by adding the additional risk factor to CHADS2 and CHA2DS2VASc. The integrated discrimination improvement (IDI) does not require risk groups and focuses on differences between integrated sensitivity and specificity for models with and without renal impairment (37). Model 1 ⫽ 1 point for renal impairment; Model 2 ⫽ 1 point for eGFR ⫽ 30–59, and 2 points for eGFR ⬍30; Model 3 ⫽ 1 point for renal impairment, and 2 points for renal impairment and eGFR ⬍30; Model 4 ⫽ 1 point for eGFR ⬍30; other abbreviations as in Tables 1 and 2.
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
the CHA2DS2-VASc score, but the improvement in the c-statistic was marginal (from 0.84 to 0.88) (30). In the present larger study, we found no significant improvement in c-statistics, and in our analyses of reclassification, there was reduced specificity despite a slight increase in sensitivity, leading to no additional predictive value for IS/TE, when renal impairment (both by eGFR and serum creatinine level) was added to the CHADS2 or CHA2DS2-VASc scores. On the basis of our analyses, renal impairment should not be added to the CHADS2 or CHA2DS2-VASc scores during routine risk stratification of AF patients for IS/TE. Indeed, renal impairment is commonly associated with the various IS risk factors listed within the CHADS2 or CHA2DS2-VASc scores; thus, the lack of an independent additive predictive value might perhaps be unsurprising. In the present study, we confirm previous observations that renal impairment, whether defined by serum creatinine level or eGFR, was associated with a more severe risk factor profile in terms of comorbidities and higher risk of IS/TE as estimated by validated risk stratification scores, compared with normal renal function (14 –16). The Cox regression analyses suggest that renal impairment is a contributing risk factor for IS/TE in AF overall, but due to small numbers of individuals with low or moderate risk (CHADS2 or CHA2DS2-VASc scores of 0 and 1, respectively), the risk of IS/TE in these subgroups could not be conclusively studied in the present population. In the recent study by Olesen et al. (15), adjusted HRs for risk of IS/TE associated with non– end-stage CKD (HR: 1.49; 95% CI: 1.38 to 1.59) and end-stage CKD requiring renal replacement therapy (HR: 1.83; 95% CI: 1.57 to 2.14) were comparable to the observations in our population. Indeed, observed event rates by renal function in the study by Olesen et al. (15) were also similar to the present study, consistent with the considerable risk of IS/TE in AF patients with renal impairment. Anticoagulation with VKA is associated with a significant risk reduction for IS/TE among patients with renal impairment, but the increased risk of bleeding in the same patient group and inconclusive data with regard to indications for oral anticoagulation (OAC) complicate the decision to start VKA in AF patients with renal impairment (15,28). As previously noted, the novel oral anticoagulant agents are predominantly renally excreted, and there are limited data with regard to their use in (severe) renal impairment (15,39). Risk stratification schemes such as CHADS2 or CHA2DS2-VASc scores are well-validated in the identification of patients at risk of IS/TE to guide the decision to initiate OAC (39), and our data suggest that renal impairment need not be added to these scores. Study limitations. The limitations of this registry include the inherent limitations of diagnostic coding and case ascertainment, particularly if an enrolled patient moved away from the area or had an outcome event in another area. Most patients with IS/TE, as far as it is identified, are likely to be seen in 1 department of our institution and not in any
Banerjee et al. Renal Impairment and Stroke Risk in AF
2085
other institution. It is possible that some patients had AF before our first recorded episode of AF. The nonrandomized cohort design does not exclude the possibility of residual confounding factors, despite statistical adjustment for several risk factors. Also, there might be clinical differences between inpatients and outpatients, which would affect the generalizability of our findings to the outpatient setting or to AF diagnosed outside the cardiology department. Inpatients usually have an acute illness or decompensation of a chronic illness that leads to hospital stay. For example, heart failure, which impacts IS/TE risk in our study, might be underrepresented in an outpatient or primary care cohort. Patients with AF seen in the cardiology department were 53% of all AF patients seen in the institution and 82% of all AF patients seen in the several medicine departments of our institution. The data with regard to VKA use only reflect baseline therapy and do not reflect any changes in prescribed therapies or adherence to therapy (which might have had multiple changes over time in a “real-world” cohort). Also, data with regard to the “time in therapeutic range” are not available for our study population. Because only baseline creatinine measurements and eGFR calculations are available, we are unable to comment definitively on changes or progression of renal impairment or the need for renal replacement therapy. For the latter, 44 (0.7%) patients had dialysis at baseline, whereas 131 (2.2%) had dialysis either at baseline or during follow-up, but a sensitivity analysis where these patients were excluded from the analysis made no difference to our observations (data not shown). Indeed, eGFR is probably the most important indicator of renal function to take into account, because OAC doses are usually lower in patients with CKD, and dose changes are more often necessary (40). In the Cox regression models and in subsequent testing of the additional value of eGFR to risk prediction of stroke/ TE, the eGFR categories of ⱖ60, 30 to 59, and ⬍30 ml/min/1.73 m2 were used to reflect the eGFR ranges most commonly used in clinical practice and to enable generalizability. Moreover, the analyses described were conducted to assess whether renal impairment (i.e., eGFR) could be added to existing risk prediction tools in a simple, useable manner. Thus, our analyses of NRI and IDI did not consider the effect of continuous increments of eGFR. In any case, eGFR was not associated with increased risk of IS/TE at 1 year, when considered as a continuous variable. The number of individuals with eGFR ⬍30 ml/min/1.73 m2 in the study population was small, and therefore the statistical power of analysis in this group might be limited. However, the proportion of individuals with low eGFR (30 ml/min/1.73 m2) in our population (5.8%) is broadly consistent with other studies (e.g., 9.95% with eGFR ⬍45 ml/min/1.73 m2 in the ATRIA [Anticoagulation and Risk Factors in Atrial Fibrillation] study (14)), suggesting that
2086
Banerjee et al. Renal Impairment and Stroke Risk in AF
our findings are likely to be representative of “real-world” clinical practice. Finally, this study had a specific focus, namely the additive impact of renal impairment on ischemic stroke risk stratification with the CHADS2 or CHA2DS2-VASc scores. It is recognized that renal impairment also increases bleeding risk (with implications for the net clinical benefit), which is beyond the focus of our present study objectives (28). Conclusions Renal impairment was not an independent predictor overall of IS/TE in patients with AF at 1 year. Adding renal impairment, whether defined by serum creatinine level or eGFR, to existing risk scoring systems for stroke/TE (CHADS2 and CHA2DS2-VASc) did not independently add to the predictive value of these scores. Reprint requests and correspondence: Prof. Gregory Y.H. Lip, University of Birmingham Centre for Cardiovascular Sciences, City Hospital, University Department of Medicine, Dudley Road, Birmingham B18 7QH, United Kingdom. E-mail: g.y.h.lip@ bham.ac.uk.
REFERENCES
1. Sarnak MJ, Levey AS, Schoolwerth AC, et al. Kidney disease as a risk factor for development of cardiovascular disease: a statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Hypertension 2003;42:1050 – 65. 2. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med 2004;351:1296 –305. 3. Weiner DE, Tighiouart H, Amin MG, et al. Chronic kidney disease as a risk factor for cardiovascular disease and all-cause mortality: a pooled analysis of community-based studies. J Am Soc Nephrol 2004;15:1307–15. 4. Coresh J, Astor B, Sarnak MJ. Evidence for increased cardiovascular disease risk in patients with chronic kidney disease. Curr Opin Nephrol Hypertens 2004;13:73– 81. 5. Couser WG, Remuzzi G, Mendis S, Tonelli M. The contribution of chronic kidney disease to the global burden of major noncommunicable diseases. Kidney Int 2011;8:1258 –70. 6. McCullough K, Sharma P, Ali T, Khan I, Smith WC, Macleod A, Black C. Measuring the population burden of chronic kidney disease: a systematic literature review of the estimated prevalence of impaired kidney function. Nephrol Dial Transplant 2012;27:1812–21. 7. Nugent RA, Fathima SF, Feigl AB, Chyung D. The burden of chronic kidney disease on developing nations: a 21st century challenge in global health. Nephron Clin Pract 2011;118:c269 –77. 8. Lloyd-Jones DM, Wang TJ, Leip EP, et al. Lifetime risk for development of atrial fibrillation: the Framingham Heart Study Circulation 2004;110:1042– 6. 9. Heeringa J, van der Kuip DA, Hofman A, et al. Prevalence, incidence and lifetime risk of atrial fibrillation: the Rotterdam Study. Eur Heart J 2006;27:949 –53. 10. Miyasaka Y, Barnes ME, Gersh BJ, et al. 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 –25. 11. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke 1991;22: 983– 8.
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87 12. Alonso A, Lopez FL, Matsushita K, et al. Chronic kidney disease is associated with the incidence of atrial fibrillation: the Atherosclerosis Risk in Communities (ARIC) study. Circulation 2011;123:2946 –53. 13. Horio T, Iwashima Y, Kamide K, et al. Chronic kidney disease as an independent risk factor for new-onset atrial fibrillation in hypertensive patients. J Hypertens 2010;28:1738 – 44. 14. Go AS, Fang MC, Udaltsova N, et al. Impact of proteinuria and glomerular filtration rate on risk of thromboembolism in atrial fibrillation: the anticoagulation and risk factors in atrial fibrillation (ATRIA) study. Circulation 2009;119:1363–9. 15. Olesen JB, Lip GY, Kamper AL, et al. Stroke and bleeding in atrial fibrillation with chronic kidney disease. N Engl J Med 2012;367: 625–35. 16. Nakagawa K, Hirai T, Takashima S, et al. Chronic kidney disease and CHADS(2) score independently predict cardiovascular events and mortality in patients with nonvalvular atrial fibrillation. Am J Cardiol 2011;107:912– 6. 17. Va´zquez E, Sa´nchez-Perales C, Lozano C, et al. Comparison of prognostic value of atrial fibrillation versus sinus rhythm in patients on long-term hemodialysis. Am J Cardiol 2003;92:868 –71. 18. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39 Suppl 1:S1–266. 19. Archibald G, Bartlett W, Brown A, et al. UK consensus conference on early chronic kidney disease. Edinburgh: Royal College of Physicians; 2007. Available at: http://www.rcpe.ac.uk/clinical-standards/ standards/UK-Consensus-Conference-on-Early-Chronic-KidneyDisease-Feb-2007.pdf. Accessed February 4, 2013. 20. National Institute for Health and Clinical Excellence (NICE). CG73 Chronic kidney disease: early identification and management of chronic kidney disease in adults in primary and secondary care. NICE clinical guidelines Issued: September 2008. Available at: http:// publications.nice.org.uk/chronic-kidney-disease-cg73/guidance. Accessed February 4, 2013. 21. 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 –70. 22. 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;137:263–72. 23. Fuster V, Rydén LE, Cannom DS, et al. 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:e101–98. 24. Camm AJ, Kirchhof P, Lip GY, et al. 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 – 429. 25. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY.A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest 2010;138:1093–100. 26. Gage BF, Yan Y, Milligan PE, et al. Clinical classification schemes for predicting hemorrhage: results from the National Registry of Atrial Fibrillation (NRAF). Am Heart J 2006;151:713–9. 27. Friberg L, Rosenqvist M, Lip GY. Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study. Eur Heart J 2012;33:1500 –10. 28. Marinigh R, Lane, DA, Lip GY. Severe renal impairment and stroke prevention in atrial fibrillation. J Am Coll Cardiol 2011;12:1339 – 48. 29. Lip GY. Chronic renal disease and stroke in atrial fibrillation: balancing the prevention of thromboembolism and bleeding risk. Europace 2011;13:145– 8. 30. Chao TF, Tsao, HM, Ambrose K, et al. Renal dysfunction and the risk of thromboembolic events in patients with atrial fibrillation after catheter ablation—the potential role beyond CHA2DS2-VASc score. Heart Rhythm 2012;9:1755– 60.
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87 31. Banerjee A, Taillandier S, Olesen JB, et al. Ejection fraction and outcomes in patients with atrial fibrillation and heart failure: the Loire Valley Atrial Fibrillation Project. Eur J Heart Fail 2012;14:295–301. 32. Couchoud C, Pozet N, Labeeuw M, Pouteil-Noble C. Screening early renal failure: cut-off values for serum creatinine as an indicator of renal impairment. Kidney Int 1999;55:1878 – 84. 33. Van Den Noortgate NJ, Janssens WH, Delanghe JR, Afschrift MB, Lameire NH. Serum cystatin C concentration compared with other markers of glomerular filtration rate in the old old. J Am Geriatr Soc 2002;50:1278 – 82. 34. Daniel JP, Chantrel F, Offner M, Moulin B, Hannedouche T. Comparison of cystatin C, creatinine and creatinine clearance vs. GFR for detection of renal failure in renal transplant patients. Ren Fail 2004;26:253–7. 35. Schuck O, Gottfriedova H, Maly J, et al. Glomerular filtration rate assessment in individuals after orthotopic liver transplantation based on serum cystatin C levels. Liver Transpl 2002;8:594 –9. 36. Mariat C, Alamartine E, Barthelemy JC, et al. Assessing renal graft function in clinical trials: can tests predicting glomerular filtration rate substitute for a reference method? Kidney Int 2004;65:289 –97. 37. Pencina MJ, D’Agostino RB Sr., D’Agostino RB Jr., Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157–72.
2087
38. Liu L, Forman S, Barton B. Fitting Cox Model Using PROC PHREG and Beyond in SAS. SAS Global Forum, Washington, DC, 2009. Statistics and Data Analysis: Paper 236 –2009. Available at: http://support.sas.com/resources/papers/proceedings09/236-2009.pdf. Accessed February 4, 2013. 39. Camm AJ, Lip GY, De Caterina R, et al. 2012 focused update of the ESC Guidelines for the management of atrial fibrillation: an update of the 2010 ESC Guidelines for the management of atrial fibrillation. Developed with the special contribution of the European Heart Rhythm Association. Eur Heart J 2012;33:2719 – 47. 40. Limdi NA, Beasley TM, Baird MF, et al. Kidney function influences warfarin responsiveness and hemorrhagic complications. J Am Soc Nephrol 2009;20:912–21. Key Words: atrial fibrillation y ischemic stroke y reclassification y risk stratification y thromboembolism. APPENDIX
For supplementary text, figures, and tables, please see the online version of this article.
Vol. 61, No. 20, 2013 ISSN 0735-1097/$36.00 http://dx.doi.org/10.1016/j.jacc.2013.02.035
Heart Rhythm Disorders
Renal Impairment and Ischemic Stroke Risk Assessment in Patients With Atrial Fibrillation The Loire Valley Atrial Fibrillation Project Amitava Banerjee, MPH, DPHIL,* Laurent Fauchier, MD, PHD,† Patrick Vourc’h, MD,‡ Christian R. Andres, MD, PHD,‡ Sophie Taillandier, MD,† Jean Michel Halimi, MD, PHD,§ Gregory Y. H. Lip, MD* Birmingham, United Kingdom; and Tours, France Objectives
This study sought to determine the risk of ischemic stroke (IS)/thromboembolism (TE) associated with renal impairment and its incremental predictive value over established risk stratification scores (congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke [CHADS2] and congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke, vascular disease, age 65 to 74 years, sex category (female) [CHA2DS2-VASc]) in patients with atrial fibrillation (AF).
Background
Risk stratification schemes for prediction of IS/TE in patients with AF are validated but do not include renal impairment.
Methods
Patients diagnosed with nonvalvular AF and available estimated glomerular filtration rate (eGFR) data in a 4-hospital institution between 2000 and 2010 were identified. The study population was stratified by renal impairment defined by serum creatinine level and by eGFR measured at time of diagnosis of AF. Independent risk factors of IS/TE (including renal impairment) were investigated in Cox regression models. The incremental predictive value of renal impairment over CHADS2 and CHA2DS2-VASc were assessed with the c-statistic, net reclassification improvement, and integrated discrimination improvement. We focused on the 1-year outcomes in our analyses.
Results
Of 8,962 eligible individuals, 5,912 (66%) had nonvalvular AF and available eGFR data. Renal impairment by both creatinine and eGFR definitions was associated with higher rates of IS/TE at 1 year, compared with normal renal function. After adjustment for CHADS2 risk factors, renal impairment did not significantly increase the risk of IS/TE at 1 year (hazard ratio: 1.06; 95% confidence interval [CI]: 0.75 to 1.49 for renal impairment; and hazard ratio: 1.09; 95% CI: 0.84 to 1.41 for eGFR). When renal impairment was added to existing risk scoring systems for stroke/TE (CHADS2 and CHA2DS2-VASc), it did not independently add to the predictive value of the scores, whether defined by serum creatinine level or eGFR. This was evident even when the analysis was confined to only those patients with at least 1 year of follow-up.
Conclusions
Renal impairment was not an independent predictor of IS/TE in patients with AF and did not significantly improve the predictive ability of the CHADS2 or CHA2DS2-VASc scores. (J Am Coll Cardiol 2013;61:2079–87) © 2013 by the American College of Cardiology Foundation
Both atrial fibrillation (AF) and chronic kidney disease (CKD) are increasingly recognized as significant burdens of morbidity and mortality of global proportions (1–11). Even mild renal impairment has long-term consequences for
cardiovascular health outcomes (5–7). Individuals with CKD are more likely to develop AF (12,13) and ischemic stroke (IS)/thromboembolism (TE) (14) than patients with normal renal function. In a recent Danish prospective study
From the *University of Birmingham Centre for Cardiovascular Sciences, City Hospital, Birmingham, United Kingdom; †Service de Cardiologie, Pôle Coeur Thorax Vasculaire, Centre Hospitalier Regional Universitaire et Faculté de Médecine, Université François Rabelais, Tours, France; ‡Service de Biochimie, Centre Hospitalier Regional Universitaire et Faculté de Médecine, Université François Rabelais, Tours, France; and the §Service de Nephrologie-Immunologie Clinique, Centre Hospitalier Regional Universitaire et Faculté de Médecine, Université François Rabelais, Tours, France. Dr. Fauchier has served as a consultant for Bayer, Medtronic, and Sanofi Aventis; and has received funding for conference travel and educational
symposia from Boehringher Ingelheim, Bayer, Medtronic, and Sanofi Aventis. Dr. Lip has served as a consultant for Bayer, Astellas, Merck, AstraZeneca, Sanofi Aventis, Biotronik, BMS/Pfizer, and Boehringher Ingelheim; and has been on the Speakers’ Bureau for Bayer, BMS/Pfizer, Boehringher Ingelheim, and Sanofi Aventis. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Halimi and Lip contributed equally to this work. Manuscript received December 11, 2012; revised manuscript received February 11, 2013, accepted February 19, 2013.
2080
Banerjee et al. Renal Impairment and Stroke Risk in AF
of 132,372 individuals with AF, 3,587 individuals had CKD, which was associated with increased risk of AF ⴝ atrial fibrillation IS/TE and bleeding (15), confirmCHADS2 ⴝ congestive ing observations of previous smaller heart failure, hypertension, studies (14,16,17). This study also age >75 years, diabetes, previous stroke showed the benefit of vitamin K antagonist (VKA) therapy on CHA2DS2-VASc ⴝ congestive heart failure, IS/TE outcomes in the setting of hypertension, age >75 CKD, although both VKA and years, diabetes, previous aspirin were associated with an stroke, vascular disease, increased risk of bleeding (15). age 65 to 74 years, sex category (female) Renal function is quantified by CI ⴝ confidence interval urinary creatinine clearance or by the estimated glomerular filtraCKD ⴝ chronic kidney disease tion rate (eGFR) (18 –20), but eGFR ⴝ estimated few studies have considered the glomerular filtration rate association between eGFR and HAS-BLED ⴝ hypertension, long-term outcomes in individuabnormal renal/liver als with AF, and existing studies function, stroke, bleeding have tended to consider renal history or predisposition, function as a dichotomous varilabile international normalized ratio, elderly able (2– 4,14). Also, patients (>65 years), drugs/alcohol with renal failure have been exconcomitantly cluded from randomized trials of HR ⴝ hazard ratio IS prevention in AF. IDI ⴝ integrated The importance of risk predicdiscrimination improvement tion tools for IS/TE (congestive IS ⴝ ischemic stroke heart failure, hypertension, age NRI ⴝ net reclassification ⱖ75 years, diabetes, previous improvement stroke [CHADS2] [21]; and OAC ⴝ oral anticoagulation congestive heart failure, hyperTE ⴝ thromboembolism tension, age ⱖ75 years, diabetes, VKA ⴝ vitamin K previous stroke, vascular disease, antagonist age 65 to 74 years, sex category (females) [CHA 2 DS 2 -VASc] [22]) in prognosis and treatment planning in AF patients is recognized by their inclusion in major international guidelines (23,24). Renal failure is included as a dichotomous variable in risk prediction tools for bleeding but not included in risk prediction tools for IS/TE (25–27), although the possibility of adding renal impairment (as the little “c”) to the CHA2DS2-VASc score has been previously proposed (28,29). In a small study of selected AF patients post-catheter ablation, renal dysfunction— defined by eGFR ⬍60 ml/min/1.73 m2— independently increased the risk of TE (hazard ratio [HR]: 6.8; 95% confidence interval [CI]: 4.2 to 12.1) (30). Therefore, better understanding of the impact of renal function on IS/TE outcomes in a more representative “real-world” population of AF patients is required. The present study represents the first analysis to consider the association between renal function, as measured by serum creatinine level or eGFR, and the risk of IS/TE events in a “real-world” population of individuals with AF, unrestricted by age or comorbidity. We also investigated the incremental predictive value of adding renal Abbreviations and Acronyms
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
function to established IS risk scores in AF (CHADS2 and CHA2DS2-VASc).
Methods Study population. The methods of the Loire Valley Atrial Fibrillation Project have been previously reported (31). Patients were followed from the first record of AF after January 1, 2000 (i.e., index date) up to the latest data collection at the time of study (December 2010) (see the Methods section in the Online Appendix). Treatment at discharge was obtained by screening hospital stay reports, and information on comorbidities was obtained from the computerized coding system. Patients were excluded from the study if there were no available data with regard to the baseline serum creatinine level at the time of first diagnosis of AF (Fig. 1). For each patient, the CHADS2 (21); CHA2DS2-VASc (22); and hypertension, abnormal renal/ liver function, stroke, bleeding history or predisposition, labile international normalized ratio, elderly (⬎65 years), drugs/alcohol concomitantly (HAS-BLED) (25) scores were calculated (Table 1). Assessment of renal function. Serum creatinine levels were taken at baseline, which was at first diagnosis of AF. Index date was the time of the first record of AF and thus the serum creatinine (and hence eGFR) measurement was from that index date. Renal impairment was defined as reported history of renal failure or baseline serum creatinine level of ⬎133 mol/l in men and ⬎115 mol/l in women (32). To convert serum creatinine from mol/l to mg/dl, the former was multiplied by a conversion factor of 88.4. Current consensus guidelines state that prediction equations have greater consistency and accuracy than serum creatinine in the assessment of GFR (18 –20,33–35). In addition, prediction equations are equivalent or better than 24-h urine creatinine clearance in all but 1 study (19,20,36). The eGFR (ml/min/1.73 m2) was calculated (see the Methods section in the Online Appendix). Outcomes. During follow-up, information on outcomes of TE (including peripheral artery embolism and transient ischemic attack), stroke (ischemic or hemorrhagic), bleeding, and all-cause mortality were recorded. In this study, the outcome of interest was IS/TE. Hemorrhagic strokes were excluded in our analyses. Statistical analysis. The study population was stratified into 3 categories according to eGFR (in ml/min/1.73 m2), corresponding to the stages of CKD: ⱖ60, 30 to 59, and ⬍30 (Fig. 1) (18 –20). Because data with regard to proteinuria were not available, stage of renal impairment could not be defined. Baseline characteristics were determined separately for the 3 eGFR strata, and differences were investigated with the chi-square test for categorical covariates and the KruskalWallis test for continuous covariates. Baseline characteristics were also determined by the presence or absence of renal impairment and then further stratification by eGFR.
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
Figure 1
2081
Study Population
Patient characteristics shown, by renal impairment and by estimated glomerular filtration rate (eGFR). VKA ⫽ vitamin K antagonist.
First, overall rates of IS/TE were calculated for patients with CHADS2 and CHA2DS2-VASc scores. Thereafter, rates were determined separately for patients with and without renal impairment and by eGFR separately and then by renal impairment with further stratification by eGFR. Second, Cox proportional-hazard regression models were constructed to investigate whether renal impairment and eGFR were independent predictors of IS/TE. The risks associated with renal impairment and eGFR were estimated in a univariate analysis, a sex- and age-adjusted analysis, an analysis adjusted for the risk factors included in the CHADS2 score, and a multivariate analysis adjusted for all baseline characteristics in Table 1, which were statistically significant (p ⬍ 0.05). All analyses were repeated by eGFR category and by combined stratification by renal impairment and eGFR. Furthermore, to test whether the results were influenced by patients initiating treatment with VKA, we performed additional analyses excluding patients at the initiation of such treatment. The value of adding renal impairment and eGFR to the CHADS2 score was evaluated by Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI), as previously described (37). Finally, we also assessed the predictive capability of CHADS2 with c-statistics estimated from Cox regression models with the method described by Liu et al. (38). In estimating the c-statistics, the CHADS2 score was analyzed as categoric risk groups (low/intermediate/high), and the effect of adding renal impairment to the score was determined.
The NRI, IDI, and the c-statistic were used to test the impact of adding renal impairment to the 2 scoring systems (CHADS2 and CHA2DS2-VASc) in 4 different statistical models by adding: 1) 1 point for renal impairment; 2) 1 point for eGFR ⫽ 30 to 59 and 2 points for eGFR ⬍30; 3) 1 point for renal impairment and 2 points for renal impairment and eGFR ⬍30; and 4) 1 point for eGFR ⬍30. These different models use the most common definitions of renal function (by eGFR and serum creatinine) in different ways to add to CHADS2 and CHA2DS2-VASc in the most user-friendly way. This approach assesses which (if any) method of adding renal impairment to either of the stroke risk scores added predictive value. We felt it was important to explore the impact of renal (dys)function in different statistical models, because renal impairment is a continuum and not just a “yes/no” phenomenon. In short, the different tested models aimed to find the most user-friendly and useful stroke risk prediction tool incorporating renal function. A 2-sided p value ⬍0.05 was considered statistically significant. All analyses were performed with SPSS statistical software (version 18.0, IBM, Chicago, Illinois), and NRI and IDI calculations were performed with R statistical software (version 2.12.2, University of Auckland, Auckland, New Zealand). Results Of 8,962 eligible individuals, 5,912 (66%) had nonvalvular AF and available serum creatinine data, allowing the eGFR
2082
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
Characteristics of Patients With AF by Renal Impairment eGFR and eGFR Table 1 Characteristics of Patients With AF by Renal and Impairment Patients With Renal Impairment eGFR (ml/min/1.73 m2)
Overall Study Population Normal Renal Function (n ⴝ 4,375)
Renal Impairment (n ⴝ 1,537)
p Value
30–59 (n ⴝ 1,196)
<30 (n ⴝ 341)
Age, yrs
68.2 ⫾ 15.7
75.8 ⫾ 10.9
⬍0.001
74.9 ⫾ 10.8
79.3 ⫾ 9.5
Female
1,777 (40.6)
419 (27.3)
⬍0.001
234 (19.6)
185 (54.3)
⬍0.001
Paroxysmal
2,635 (60.2)
760 (49.4)
⬍0.001
583 (48.7)
177 (51.9)
0.49
Permanent
1501 (34.3)
688 (44.8)
541 (45.2)
147 (43.1)
Persistent
239 (5.5)
89 (5.8)
72 (6.0)
17 (5.0)
p Value
Type of AF
Comorbidities 1,780 (40.7)
757 (49.3)
⬍0.001
548 (45.8)
209 (61.3)
Diabetes
653 (14.9)
315 (20.5)
⬍0.001
220 (18.4)
95 (27.9)
0.001
Previous stroke
315 (7.2)
133 (8.7)
0.06
94 (7.9)
39 (11.4)
0.11
Coronary artery disease
1,215 (27.8)
631 (41.1)
⬍0.001
503 (42.1)
128 (37.5)
0.16
Any vascular disease
1,336 (30.5)
698 (45.4)
⬍0.001
549 (45.9)
149 (43.7)
0.33
Heart failure
2,083 (47.6)
1,015 (66.0)
⬍0.001
760 (63.5)
255 (74.8)
⬍0.001
Hypertension
Liver impairment Dyslipidemia
⬍0.001
14 (0.3)
4 (0.3)
0.72
3 (0.3)
1 (0.3)
0.99
821 (18.8)
327 (21.3)
0.03
254 (21.2)
73 (21.4)
0.76
Smoking
531 (12.1)
213 (13.9)
0.08
165 (13.8)
48 (14.1)
0.85
Pacemaker/ICD
689 (15.7)
349 (22.7)
⬍0.001
277 (23.2)
72 (21.1)
0.54
Bleeding risk factors 192 (4.4)
91 (5.9)
0.02
64 (5.4)
27 (7.9)
0.20
Labile INR
Previous bleeding
80 (1.8)
24 (1.6)
0.49
15 (1.3)
9 (2.6)
0.19
Anemia
21 (0.5)
16 (1.0)
0.02
11 (0.9)
5 (1.5)
0.68
NSAIDs
6 (0.1)
2 (0.1)
0.95
2 (0.2)
0 (0.0)
0.75
Drugs
717 (16.4)
365 (23.7)
⬍0.001
290 (24.2)
75 (22.0)
0.50
Cancer
72 (1.6)
32 (2.1)
0.26
22 (1.8)
10 (2.9)
0.45
Excessive risk of falls
40 (0.9)
23 (1.5)
0.06
15 (1.3)
8 (2.3)
0.34
4 (0.1)
1 (0.1)
0.76
1 (0.1)
0 (0.0)
0.87 ⬍0.001
Thrombocytopenia Antithrombotic agents Vitamin K antagonist
2,295 (52.5)
796 (51.8)
0.58
654 (54.7)
142 (46.9)
Antiplatelet
1,349 (30.8)
525 (34.2)
0.002
408 (34.1)
117 (39.4)
0.52
Any antithrombotic
3,238 (74.0)
1127 (73.3)
0.21
892 (74.6)
235 (78.1)
0.001
Heart failure therapy ACEI
779 (17.8)
350 (22.8)
0.06
283 (23.7)
67 (19.6)
0.02
Beta-blocker
981 (22.4)
425 (27.7)
0.16
344 (28.8)
81 (23.8)
0.004
Digoxin
563 (12.9)
206 (13.4)
0.11
166 (13.9)
40 (11.7)
0.14
Diuretic
807 (18.4)
467 (30.4)
⬍0.001
348 (29.1)
119 (34.9)
0.07
1,222 (27.9)
443 (28.8)
0.52
347 (29.0)
96 (28.2)
0.49
157 (3.6)
70 (4.6)
0.01
52 (4.3)
18 (5.3)
0.12
⬍0.001
⬍0.001
Antiarrhythmic agent Calcium channel blocker CHADS2 Low (score ⫽ 0)
1,084 (24.8)
127 (8.3)
122 (10.2)
5 (1.5)
Intermediate (score ⫽ 1)
1,119 (25.6)
373 (24.3)
318 (26.6)
55 (16.1)
High (score ⱖ2)
2,172 (49.6)
1037 (67.5)
756 (63.2)
281 (82.4)
CHA2DS2-VASc Low (score ⫽ 0)
475 (10.9)
38 (2.5)
37 (3.1)
1 (0.3)
Intermediate (score ⫽ 1)
682 (15.6)
91 (5.9)
86 (7.2)
5 (1.5)
3,218 (73.6)
1408 (91.6)
1073 (89.7)
335 (98.2)
Low (score ⫽ 0)
1,030 (23.5)
101 (6.6)
95 (7.9)
6 (1.8)
Moderate (score ⫽ 1–2)
1,628 (37.2)
544 (35.4)
480 (40.1)
64 (18.8)
High (score ⱖ3)
1,717 (39.2)
892 (58.0)
621 (51.9)
136 (39.9)
High (score ⱖ2)
⬍0.001
⬍0.001
HAS-BLED ⬍0.001
⬍0.001
Values are mean ⫾ SD or n (%). According to the congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke (CHADS2) and congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke, vascular disease, age 65 to 74 years, sex category (female) (CHA2DS2-VASc) scores, patients with a score of 0 on either schema were considered as “low risk,” 1 as “intermediate risk,” and ⱖ2 as “high risk” of stroke and thromboembolism. The Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile International Normalized Ratio, Elderly (⬎65 years), Drugs/alcohol concomitantly (HAS-BLED) score is a validated scoring system for bleeding risk stratification in atrial fibrillation (AF) patients (25). Patients with HAS-BLED score of 0 to 2 were deemed to have “low-moderate” bleeding risk, and those with HAS-BLED score of ⱖ3 were classified as “high” bleeding risk. CHADS2 (1 point each for heart failure, hypertension, age ⱖ75 and diabetes, and 2 points for prior stroke or thromboembolism); CHA2DS2-VASc (1 point for heart failure, hypertension, diabetes, vascular disease, age 65 to 74, and female sex; 2 points for prior stroke or thromboembolism and age ⱖ75). ACEI ⫽ angiotensin-converting enzyme inhibitor; eGFR ⫽ estimated glomerular filtration rate; ICD ⫽ implantable cardiac defibrillator; INR ⫽ international normalized ratio.
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
Patients Rate of Stroke/TE and in Patients per 100 With Person-Years CHADS in Allⴝ 0 With and Without Renal Impairment Rate of Stroke/TE per 100 Person-Years in All 2 Score Table 2 Patients and in Patients With CHADS2 Score ⴝ 0 With and Without Renal Impairment 1-Yr Follow-Up
Person-Yrs
Stroke/TE Events
Stroke/TE Rate (CI)
Overall (whole cohort) Total
3,669
171
4.7 (3.5–6.2)
eGFR ⱖ60
1,863
64
3.4 (2.4–4.8)
eGFR ⫽ 30–59
1,610
92
5.7 (4.2–7.8)
196
15
7.7 (4.3–13.6)
2,738
119
4.4 (3.2–5.9)
Renal impairment
931
52
5.6 (3.9–8.1)
eGFR ⫽ 30–59
735
37
5.0 (3.3–7.6)
eGFR ⬍30
196
15
7.7 (4.3–13.6)
eGFR ⬍30 Normal renal function
CHADS2 ⫽ 0 Total
688
11
1.6 (0.8–3.0)
eGFR ⱖ60
521
9
1.7 (0.9–3.4)
eGFR ⫽ 30–59
163
2
1.2 (0.3–5.0)
4
0
—
614
10
1.6 (0.8–3.2)
Renal impairment
74
1
1.4 (0.2–9.8)
eGFR ⫽ 30–59
70
1
1.4 (0.2–10.4)
4
0
—
eGFR ⬍30 Normal renal function
eGFR ⬍30
CI ⫽ confidence interval; TE ⫽ thromboembolism; other abbreviations as in Table 1.
to be calculated (Fig. 1). Of the cohort, 3,944 individuals had ⱕ1 year of follow-up, whereas 1,444 individuals had 2 to 5 years of follow-up, and 524 had 6 to 10 years of follow-up. Thus, 14,499 patient-years of follow-up were included in the analysis, with mean follow-up of 2.45 ⫾ 3.56 years. We focused on the 1-year outcomes in our analyses. Of the cohort, 1,537 (26%) individuals had renal impairment, and of these, 341 (5.8%) had eGFR ⬍30 ml/min/ 1.73 m2 (Fig. 1). Online Figure 1 illustrates the study population by eGFR alone. Baseline characteristics are shown in Table 1 by renal impairment and eGFR category and in Online Table 1 by eGFR category. Individuals with renal impairment were older, more likely to be male, less likely to have paroxysmal AF, more likely to have comorbidities, and at higher risk of stroke/TE (assessed by CHADS2 and CHA2DS2-VASc scores) and bleeding (assessed by HAS-BLED score) (Table 1). Similar trends were present when those with renal impairment were further stratified by eGFR (Table 1) and when the overall population was stratified by eGFR (Online Table 1). Of 434 IS/TE events, 171 (39.4%) occurred within the first year. Table 2 shows the rates of stroke/TE per 100 person-years according to the presence or absence of renal impairment in patients with CHADS2 score ⫽ 0 and in the overall population. Subjects with normal renal function had lower event rates for IS/TE at 1 year. Patients with renal impairment had a higher rate of stroke/TE, and patients with eGFR ⬍30 ml/min/1.73 m2 had a higher rate than those patients with eGFR 30 to 59 ml/min/1.73 m2 in the overall population. Due to the small number of individuals
2083
with CHADS2 score ⫽ 0, there were no events in the eGFR ⬍30 ml/min/1.73 m2 category, and the rate of IS/TE could not be calculated for this particular category, and CIs for event rates in other subgroups of renal function were wide. Rates of stroke/TE were 3.3% and 7% at 1 year in individuals with eGFR ⱖ60 ml/min/1.73 m2 and with eGFR ⬍30 ml/min/1.73 m2, respectively. The corresponding rates of all-cause mortality at 1 year were 4.2% and 17.6%, respectively (Online Fig. 2). Online Table 2 displays the rates in patients with a CHA2DS2-VASc score ⫽ 0 to 1. Only 91 patients with renal impairment had a CHA2DS2-VASc score ⫽ 1 and only 2 events occurred in this subgroup. No statistically significant differences in IS/TE rates were observed between individuals when this subpopulation was stratified by renal function. Table 3 shows the results from the Cox regression analyses, showing the association of risk factors and risk of IS/TE. As a categoric variable (but not as a continuous variable), eGFR was an independent predictor of IS/TE in AF at 1 year follow-up—after adjustment for age, sex, and CHADS2 risk factors— but not for baseline characteristics. Renal impairment did not significantly increase the risk of IS/TE in univariate analyses or after adjustment for CHADS2 risk factors, age, sex, or baseline characteristics at 1 year. When considered as a continuous variable, eGFR was also not associated with an increased risk of IS/TE at 1 year. Online Table 3 shows analogous results from Cox regression analyses after excluding patients on a regimen of vitamin K antagonists at baseline (n ⫽ 3,592; 60.8%). Renal Results Renal Impairment FromRenal Cox and Regression Risk of Analyses Stroke/TE: Impairment and Risk of Stroke/TE: Table 3 Results From Cox Regression Analyses 1-Yr Follow-Up HR (95% CI) Univariate analysis Renal impairment
1.05 (0.76–1.46)
eGFR (categoric variable)
1.15 (0.90–1.47)
Adjusted for sex and age Renal impairment
1.03 (0.73–1.43)
eGFR (categoric variable)
1.07 (0.82–1.40)
Female
1.16 (0.84–1.61)
Age/5-yr increase
1.04 (0.99–1.02)
Adjusted for CHADS2 risk factors Renal impairment
1.06 (0.75–1.49)
eGFR (categoric variable)
1.09 (0.84–1.41)
Heart failure
0.91 (0.65–1.27)
Hypertension
1.22 (0.88–1.68)
Age ⱖ 75 yrs
1.25 (0.77–2.02)
Diabetes mellitus
1.22 (0.82–1.80)
Previous stroke/TE
0.88 (0.62–1.26)
Adjusted for baseline characteristics* Renal impairment
1.09 (0.77–1.55)
eGFR (categoric variable)
1.08 (0.83–1.41)
*Adjusted for Table 1 risk factors that were statistically significant (p ⬍ 0.05), and including age as a continuous covariate, the result is only displayed for renal impairment. HR ⫽ hazard ratio; other abbreviations as in Tables 1 and 2.
2084
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
In the overall study population, the c-statistic for the CHADS2 (HR: 0.64; 95% CI: 0.61 to 0.67) and CHA2DS2-VASC (HR: 0.64; 95% CI: 0.62 to 0.67) were similar. There was no statistically significant improvement in either the CHADS2 or the CHA2DS2-VASc scoring systems by the addition of renal function, regardless of the 4 models used to add renal function to the risk scores, or method of estimating reclassification (i.e., NRI or IDI) (Table 4). This was still evident even when the analysis was confined to only those patients with at least 1 year of follow-up (Online Table 4). Discussion
Figure 2
Risk of Stroke/Thromboembolism With Renal Impairment
Results from Cox regression analyses. CHADS2 ⫽ congestive heart failure, hypertension, age ⱖ75 years, diabetes, previous stroke; eGFR ⫽ estimated glomerular filtration rate.
impairment was not an independent predictor for IS/TE on univariate analysis (HR: 1.51; 95% CI: 0.93 to 2.44), after adjusting for age/sex (HR: 1.16; 95% CI: 0.71 to 1.90), CHADS2 risk factors (HR: 1.20; 95% CI: 0.73 to 1.96), or baseline characteristics (HR: 0.87; 95% CI: 0.46 to 1.65) at 1 year of follow-up. The Forest plots in Figure 2 illustrate the results from Cox regression analyses evaluating differences in IS/TE risk according to renal function, confirming that renal impairment and eGFR did not significantly increase the risk of IS/TE after univariate or multivariate models at 1 year. Table 4 illustrates the impact of adding renal impairment to the CHADS2 and CHA2DS2-VASc scores, as measured by NRI, IDI, and c-statistic. Online Table 4 shows the same analysis confined to only those patients with at least 1 year of follow-up (294 IS/TE events in 2,663 individuals over the study period).
In this study amongst a large “real-world” cohort of individuals with nonvalvular AF, we have shown for the first time that adding renal impairment to existing stroke risk stratification systems for IS/TE (CHADS2 and CHA2DS2VASc) did not independently add to the predictive value of these scores at 1 year of follow-up, whether it was defined by serum creatinine level or the eGFR or whether the analysis was confined to only those patients with at least 1 year of follow-up. Second, we show that renal impairment by both creatinine and eGFR definitions was associated with higher crude rates of stroke/TE at 1 year compared with those with normal renal function, but renal impairment was not independently associated with increased risk of stroke/TE at 1 year. Although it has been well-recognized that renal impairment is a risk factor for IS/TE in AF patients, renal impairment has not been included in the CHADS2 or CHA2DS2-VASc scores, due to lack of large prospective cohort data validating the additive value of renal impairment to these scores. Trial datasets would not answer this question, because severe renal impairment was often an exclusion criteria for clinical trials of IS prevention in AF. One highly selected cohort of AF patients undergoing ablation did suggest an additive value of renal dysfunction to
NRI and4IDI by Renal Impairment the CHADS andCHADS CHA2DS Scores 2VASc Table NRIAdding and IDI by Adding Renal to Impairment to2 the CHA2DS 2 and 2VASc Scores Patients Reclassified (%)
NRI
IDI
Patients With Event (n ⴝ 434)
Patients Without Event (n ⴝ 5,478)
NRI
SE
Z
p Value
IDI
AUC*
p Value
CHADS2 Model 1
31 (7.1)
469 (8.6)
⫺0.015
0.011
⫺0.03
0.49
0.0001
⫺0.004
0.41
CHADS2 Model 2
54 (12.4)
957 (17.5)
⫺0.051
0.010
0.27
0.61
⫺0.0001
⫺0.001
0.57
CHADS2 Model 3
31 (7.1)
469 (8.6)
⫺0.016
0.011
⫺0.08
0.47
⫺0.0001
⫺0.007
0.40
CHADS2 Model 4
5 (1.2)
55 (1.0)
0.002
0.011
0.54
0.71
0.0001
⬍0.0001
0.44
CHA2DS2VASc Model 1
2 (0.5)
127 (2.3)
⫺0.018
0.008
⫺0.90
0.18
⫺0.0001
⫺0.005
0.44
CHA2DS2VASc Model 2
2 (0.5)
238 (4.3)
⫺0.039
0.008
⫺0.80
0.21
0.0002
⫺0.014
0.30
CHA2DS2VASc Model 3
2 (0.5)
127 (2.3)
⫺0.018
0.008
⫺0.90
0.18
⫺0.0001
⫺0.005
0.44
CHA2DS2VASc Model 4
0 (0.0)
5 (0.1)
⫺0.001
0.009
⫺0.05
0.48
0.0000
⬍0.0001
0.25
The net reclassification improvement (NRI) focuses on reclassification tables constructed separately for patients with and without event and quantifies the correct movement in CHADS2 score by adding renal impairment (i.e., increased CHADS2 score for events by adding renal impairment counts positively, and increased CHADS2 score for non-events counts negatively). *Improvement in the area under the receiver operator characteristic curve (AUC) (c-statistic) by adding the additional risk factor to CHADS2 and CHA2DS2VASc. The integrated discrimination improvement (IDI) does not require risk groups and focuses on differences between integrated sensitivity and specificity for models with and without renal impairment (37). Model 1 ⫽ 1 point for renal impairment; Model 2 ⫽ 1 point for eGFR ⫽ 30–59, and 2 points for eGFR ⬍30; Model 3 ⫽ 1 point for renal impairment, and 2 points for renal impairment and eGFR ⬍30; Model 4 ⫽ 1 point for eGFR ⬍30; other abbreviations as in Tables 1 and 2.
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87
the CHA2DS2-VASc score, but the improvement in the c-statistic was marginal (from 0.84 to 0.88) (30). In the present larger study, we found no significant improvement in c-statistics, and in our analyses of reclassification, there was reduced specificity despite a slight increase in sensitivity, leading to no additional predictive value for IS/TE, when renal impairment (both by eGFR and serum creatinine level) was added to the CHADS2 or CHA2DS2-VASc scores. On the basis of our analyses, renal impairment should not be added to the CHADS2 or CHA2DS2-VASc scores during routine risk stratification of AF patients for IS/TE. Indeed, renal impairment is commonly associated with the various IS risk factors listed within the CHADS2 or CHA2DS2-VASc scores; thus, the lack of an independent additive predictive value might perhaps be unsurprising. In the present study, we confirm previous observations that renal impairment, whether defined by serum creatinine level or eGFR, was associated with a more severe risk factor profile in terms of comorbidities and higher risk of IS/TE as estimated by validated risk stratification scores, compared with normal renal function (14 –16). The Cox regression analyses suggest that renal impairment is a contributing risk factor for IS/TE in AF overall, but due to small numbers of individuals with low or moderate risk (CHADS2 or CHA2DS2-VASc scores of 0 and 1, respectively), the risk of IS/TE in these subgroups could not be conclusively studied in the present population. In the recent study by Olesen et al. (15), adjusted HRs for risk of IS/TE associated with non– end-stage CKD (HR: 1.49; 95% CI: 1.38 to 1.59) and end-stage CKD requiring renal replacement therapy (HR: 1.83; 95% CI: 1.57 to 2.14) were comparable to the observations in our population. Indeed, observed event rates by renal function in the study by Olesen et al. (15) were also similar to the present study, consistent with the considerable risk of IS/TE in AF patients with renal impairment. Anticoagulation with VKA is associated with a significant risk reduction for IS/TE among patients with renal impairment, but the increased risk of bleeding in the same patient group and inconclusive data with regard to indications for oral anticoagulation (OAC) complicate the decision to start VKA in AF patients with renal impairment (15,28). As previously noted, the novel oral anticoagulant agents are predominantly renally excreted, and there are limited data with regard to their use in (severe) renal impairment (15,39). Risk stratification schemes such as CHADS2 or CHA2DS2-VASc scores are well-validated in the identification of patients at risk of IS/TE to guide the decision to initiate OAC (39), and our data suggest that renal impairment need not be added to these scores. Study limitations. The limitations of this registry include the inherent limitations of diagnostic coding and case ascertainment, particularly if an enrolled patient moved away from the area or had an outcome event in another area. Most patients with IS/TE, as far as it is identified, are likely to be seen in 1 department of our institution and not in any
Banerjee et al. Renal Impairment and Stroke Risk in AF
2085
other institution. It is possible that some patients had AF before our first recorded episode of AF. The nonrandomized cohort design does not exclude the possibility of residual confounding factors, despite statistical adjustment for several risk factors. Also, there might be clinical differences between inpatients and outpatients, which would affect the generalizability of our findings to the outpatient setting or to AF diagnosed outside the cardiology department. Inpatients usually have an acute illness or decompensation of a chronic illness that leads to hospital stay. For example, heart failure, which impacts IS/TE risk in our study, might be underrepresented in an outpatient or primary care cohort. Patients with AF seen in the cardiology department were 53% of all AF patients seen in the institution and 82% of all AF patients seen in the several medicine departments of our institution. The data with regard to VKA use only reflect baseline therapy and do not reflect any changes in prescribed therapies or adherence to therapy (which might have had multiple changes over time in a “real-world” cohort). Also, data with regard to the “time in therapeutic range” are not available for our study population. Because only baseline creatinine measurements and eGFR calculations are available, we are unable to comment definitively on changes or progression of renal impairment or the need for renal replacement therapy. For the latter, 44 (0.7%) patients had dialysis at baseline, whereas 131 (2.2%) had dialysis either at baseline or during follow-up, but a sensitivity analysis where these patients were excluded from the analysis made no difference to our observations (data not shown). Indeed, eGFR is probably the most important indicator of renal function to take into account, because OAC doses are usually lower in patients with CKD, and dose changes are more often necessary (40). In the Cox regression models and in subsequent testing of the additional value of eGFR to risk prediction of stroke/ TE, the eGFR categories of ⱖ60, 30 to 59, and ⬍30 ml/min/1.73 m2 were used to reflect the eGFR ranges most commonly used in clinical practice and to enable generalizability. Moreover, the analyses described were conducted to assess whether renal impairment (i.e., eGFR) could be added to existing risk prediction tools in a simple, useable manner. Thus, our analyses of NRI and IDI did not consider the effect of continuous increments of eGFR. In any case, eGFR was not associated with increased risk of IS/TE at 1 year, when considered as a continuous variable. The number of individuals with eGFR ⬍30 ml/min/1.73 m2 in the study population was small, and therefore the statistical power of analysis in this group might be limited. However, the proportion of individuals with low eGFR (30 ml/min/1.73 m2) in our population (5.8%) is broadly consistent with other studies (e.g., 9.95% with eGFR ⬍45 ml/min/1.73 m2 in the ATRIA [Anticoagulation and Risk Factors in Atrial Fibrillation] study (14)), suggesting that
2086
Banerjee et al. Renal Impairment and Stroke Risk in AF
our findings are likely to be representative of “real-world” clinical practice. Finally, this study had a specific focus, namely the additive impact of renal impairment on ischemic stroke risk stratification with the CHADS2 or CHA2DS2-VASc scores. It is recognized that renal impairment also increases bleeding risk (with implications for the net clinical benefit), which is beyond the focus of our present study objectives (28). Conclusions Renal impairment was not an independent predictor overall of IS/TE in patients with AF at 1 year. Adding renal impairment, whether defined by serum creatinine level or eGFR, to existing risk scoring systems for stroke/TE (CHADS2 and CHA2DS2-VASc) did not independently add to the predictive value of these scores. Reprint requests and correspondence: Prof. Gregory Y.H. Lip, University of Birmingham Centre for Cardiovascular Sciences, City Hospital, University Department of Medicine, Dudley Road, Birmingham B18 7QH, United Kingdom. E-mail: g.y.h.lip@ bham.ac.uk.
REFERENCES
1. Sarnak MJ, Levey AS, Schoolwerth AC, et al. Kidney disease as a risk factor for development of cardiovascular disease: a statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Hypertension 2003;42:1050 – 65. 2. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med 2004;351:1296 –305. 3. Weiner DE, Tighiouart H, Amin MG, et al. Chronic kidney disease as a risk factor for cardiovascular disease and all-cause mortality: a pooled analysis of community-based studies. J Am Soc Nephrol 2004;15:1307–15. 4. Coresh J, Astor B, Sarnak MJ. Evidence for increased cardiovascular disease risk in patients with chronic kidney disease. Curr Opin Nephrol Hypertens 2004;13:73– 81. 5. Couser WG, Remuzzi G, Mendis S, Tonelli M. The contribution of chronic kidney disease to the global burden of major noncommunicable diseases. Kidney Int 2011;8:1258 –70. 6. McCullough K, Sharma P, Ali T, Khan I, Smith WC, Macleod A, Black C. Measuring the population burden of chronic kidney disease: a systematic literature review of the estimated prevalence of impaired kidney function. Nephrol Dial Transplant 2012;27:1812–21. 7. Nugent RA, Fathima SF, Feigl AB, Chyung D. The burden of chronic kidney disease on developing nations: a 21st century challenge in global health. Nephron Clin Pract 2011;118:c269 –77. 8. Lloyd-Jones DM, Wang TJ, Leip EP, et al. Lifetime risk for development of atrial fibrillation: the Framingham Heart Study Circulation 2004;110:1042– 6. 9. Heeringa J, van der Kuip DA, Hofman A, et al. Prevalence, incidence and lifetime risk of atrial fibrillation: the Rotterdam Study. Eur Heart J 2006;27:949 –53. 10. Miyasaka Y, Barnes ME, Gersh BJ, et al. 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 –25. 11. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke 1991;22: 983– 8.
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87 12. Alonso A, Lopez FL, Matsushita K, et al. Chronic kidney disease is associated with the incidence of atrial fibrillation: the Atherosclerosis Risk in Communities (ARIC) study. Circulation 2011;123:2946 –53. 13. Horio T, Iwashima Y, Kamide K, et al. Chronic kidney disease as an independent risk factor for new-onset atrial fibrillation in hypertensive patients. J Hypertens 2010;28:1738 – 44. 14. Go AS, Fang MC, Udaltsova N, et al. Impact of proteinuria and glomerular filtration rate on risk of thromboembolism in atrial fibrillation: the anticoagulation and risk factors in atrial fibrillation (ATRIA) study. Circulation 2009;119:1363–9. 15. Olesen JB, Lip GY, Kamper AL, et al. Stroke and bleeding in atrial fibrillation with chronic kidney disease. N Engl J Med 2012;367: 625–35. 16. Nakagawa K, Hirai T, Takashima S, et al. Chronic kidney disease and CHADS(2) score independently predict cardiovascular events and mortality in patients with nonvalvular atrial fibrillation. Am J Cardiol 2011;107:912– 6. 17. Va´zquez E, Sa´nchez-Perales C, Lozano C, et al. Comparison of prognostic value of atrial fibrillation versus sinus rhythm in patients on long-term hemodialysis. Am J Cardiol 2003;92:868 –71. 18. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39 Suppl 1:S1–266. 19. Archibald G, Bartlett W, Brown A, et al. UK consensus conference on early chronic kidney disease. Edinburgh: Royal College of Physicians; 2007. Available at: http://www.rcpe.ac.uk/clinical-standards/ standards/UK-Consensus-Conference-on-Early-Chronic-KidneyDisease-Feb-2007.pdf. Accessed February 4, 2013. 20. National Institute for Health and Clinical Excellence (NICE). CG73 Chronic kidney disease: early identification and management of chronic kidney disease in adults in primary and secondary care. NICE clinical guidelines Issued: September 2008. Available at: http:// publications.nice.org.uk/chronic-kidney-disease-cg73/guidance. Accessed February 4, 2013. 21. 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 –70. 22. 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;137:263–72. 23. Fuster V, Rydén LE, Cannom DS, et al. 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:e101–98. 24. Camm AJ, Kirchhof P, Lip GY, et al. 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 – 429. 25. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY.A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest 2010;138:1093–100. 26. Gage BF, Yan Y, Milligan PE, et al. Clinical classification schemes for predicting hemorrhage: results from the National Registry of Atrial Fibrillation (NRAF). Am Heart J 2006;151:713–9. 27. Friberg L, Rosenqvist M, Lip GY. Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study. Eur Heart J 2012;33:1500 –10. 28. Marinigh R, Lane, DA, Lip GY. Severe renal impairment and stroke prevention in atrial fibrillation. J Am Coll Cardiol 2011;12:1339 – 48. 29. Lip GY. Chronic renal disease and stroke in atrial fibrillation: balancing the prevention of thromboembolism and bleeding risk. Europace 2011;13:145– 8. 30. Chao TF, Tsao, HM, Ambrose K, et al. Renal dysfunction and the risk of thromboembolic events in patients with atrial fibrillation after catheter ablation—the potential role beyond CHA2DS2-VASc score. Heart Rhythm 2012;9:1755– 60.
Banerjee et al. Renal Impairment and Stroke Risk in AF
JACC Vol. 61, No. 20, 2013 May 21, 2013:2079–87 31. Banerjee A, Taillandier S, Olesen JB, et al. Ejection fraction and outcomes in patients with atrial fibrillation and heart failure: the Loire Valley Atrial Fibrillation Project. Eur J Heart Fail 2012;14:295–301. 32. Couchoud C, Pozet N, Labeeuw M, Pouteil-Noble C. Screening early renal failure: cut-off values for serum creatinine as an indicator of renal impairment. Kidney Int 1999;55:1878 – 84. 33. Van Den Noortgate NJ, Janssens WH, Delanghe JR, Afschrift MB, Lameire NH. Serum cystatin C concentration compared with other markers of glomerular filtration rate in the old old. J Am Geriatr Soc 2002;50:1278 – 82. 34. Daniel JP, Chantrel F, Offner M, Moulin B, Hannedouche T. Comparison of cystatin C, creatinine and creatinine clearance vs. GFR for detection of renal failure in renal transplant patients. Ren Fail 2004;26:253–7. 35. Schuck O, Gottfriedova H, Maly J, et al. Glomerular filtration rate assessment in individuals after orthotopic liver transplantation based on serum cystatin C levels. Liver Transpl 2002;8:594 –9. 36. Mariat C, Alamartine E, Barthelemy JC, et al. Assessing renal graft function in clinical trials: can tests predicting glomerular filtration rate substitute for a reference method? Kidney Int 2004;65:289 –97. 37. Pencina MJ, D’Agostino RB Sr., D’Agostino RB Jr., Vasan RS. Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008;27:157–72.
2087
38. Liu L, Forman S, Barton B. Fitting Cox Model Using PROC PHREG and Beyond in SAS. SAS Global Forum, Washington, DC, 2009. Statistics and Data Analysis: Paper 236 –2009. Available at: http://support.sas.com/resources/papers/proceedings09/236-2009.pdf. Accessed February 4, 2013. 39. Camm AJ, Lip GY, De Caterina R, et al. 2012 focused update of the ESC Guidelines for the management of atrial fibrillation: an update of the 2010 ESC Guidelines for the management of atrial fibrillation. Developed with the special contribution of the European Heart Rhythm Association. Eur Heart J 2012;33:2719 – 47. 40. Limdi NA, Beasley TM, Baird MF, et al. Kidney function influences warfarin responsiveness and hemorrhagic complications. J Am Soc Nephrol 2009;20:912–21. Key Words: atrial fibrillation y ischemic stroke y reclassification y risk stratification y thromboembolism. APPENDIX
For supplementary text, figures, and tables, please see the online version of this article.