A nationwide cohort study of the role of CHADS₂ score in predicting lower extremity amputation and death among patients with peripheral arterial occlusive disease

Abstract

Objectives: The current study aimed to explore whether the CHADS₂ score was predictive of lower extremity amputation (LEA) and death in people with peripheral arterial occlusive disease (PAOD).

Methods: This nationwide cohort came from Taiwan, with 16,888 PAOD patients, from 2000 through 2011, extracted from the Longitudinal Health Insurance Database 2000. Cox proportional hazard regression models were employed to identify the LEA and mortality risk according to CHADS₂ score. The discriminatory properties of the score in predicting the outcomes were quantified by the area under the receiver operating characteristic curve (AUROC) and the Cox C-index.

Results: The AUROC of the CHADS₂ score in predicting LEA and death were 0.75 (95% CI = 0.73–0.77) and 0.70 (95% CI = 0.69–0.71), respectively. The CHADS₂ score had an acceptable stratification capacity for LEA (C-index = 0.79) and death (C-index = 0.76) based on Cox-regression analysis.

Conclusions: This study correlates the CHADS₂ score with risk of developing LEA and death in patients with PAOD. The acceptable discriminative power of the score diversifies its predictive role in this population.

Keywords:

• CHADS₂ score
• death
• lower extremity amputation
• peripheral arterial occlusive disease

Introduction

The burden of atherosclerosis-related diseases is huge worldwide; indeed, three common atherosclerosis diseases, including coronary artery disease (CAD), cerebral vascular disease (CVD), and peripheral arterial occlusive disease (PAOD) constitute the atherosclerotic cardiovascular disease (ASCAD) [1].

Emerging role of PAOD has been noted recently; indeed, approximately two-thirds PAOD patients have some degree of CAD or CVD, highlighting the importance of PAOD surveillance [2,3]. Once PAOD was left undiagnosed and untreated, several sequelae might occur; among them, loss of lower limb gain more attention due to the deleterious biomedical and psychosocial effect [2,3]. Hence, to avoid the unexpected and unnecessary limb loss in PAOD patients, to construct a simply stratification tool for practical clinical use is urgently needed.

Expanded indications of CHADS₂ score for predicting adverse outcomes in different groups have been reported although this particular score was originally designed for the evaluation of stroke risk among people with established AF [4–8]. The current study was conducted to investigate whether the CHADS₂ score, a real-made product, was predictive of incident lower limb amputation in people with PAOD. Additionally, the other outcome of all-cause mortality was also included for analysis.

Methods

Data source

Data were sourced from the Longitudinal Health Insurance Database (LHID) 2000. For research purpose, the National Health Research Institutes (NHRI) compiles all medical claims in the National Health Insurance (NHI) program and releases the database annually to the public [9]. The details of the NHI program and LHID 2000 were reported elsewhere [10,11]. Diagnosis was made according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes. This study was approved by the Institutional Review Board of China Medical University (CMUH-104-REC2-115).

Sampled participants

We selected patients aged ≥18 years with newly diagnosed PAOD (ICD-9-CM codes 440.2, 440.3, 440.8, 440.9, 443, 444.0, 444.22, 444.8, 447.8, 447.9), between 2000 and 2011, as the study cohort. PAOD patients with a previous diagnosis of lower extremity amputation (LEA) (ICD-9 procedure codes 84.1) were excluded. The PAOD diagnosis date was defined as the index date. The CHADS₂ score was calculated for each patient by assigning 1 point each for age ≧75 years, a history of heart failure, hypertension, and diabetes mellitus; 2 points each for a history of an ischemic stroke/transient ischemic attack (TIA) [12–15]. Other baseline comorbidities, such as atrial fibrillation (AF) (ICD-9-CM code 427.31), hyperlipidemia (ICD-9-CM code 272), chronic obstructive pulmonary disease (COPD) (ICD-9-CM codes 491, 492, 496), chronic kidney disease (CKD) (ICD-9-CM codes 580-589), and gout (ICD-9-CM code 274) were also included in the current study.

Outcomes

All study patients were followed from the index date to the occurrence of LEA, death, withdrawal from the NHI program, or the end of 2011 (whichever came first).

Statistical analysis

To characterize the PAOD cohort, we presented number and percentage for category variables, and mean and corresponding standard deviation (SD) for continuous variables. We calculated the incidence density rate of LEA and mortality according to CHADS₂ score. Univariable and multivariable Cox proportional hazard regression models were employed to explore the risk of LEA and death according to CHADS₂ score, shown as hazard ratios (HRs) with 95% confidence intervals (CIs). Cumulative incidence curves of LEA and mortality stratified by CHADS₂ score were measured by Kaplan–Meier method. We applied log-rank test to assess the difference between curves. The predictive ability of CHADS₂ score for LEA and mortality in the study cohort was assessed using area under the receiver operating characteristic (AUROC) curve. The C-index was also estimated based on Cox-regression analysis (Harrell C) to quantify the discriminatory properties of the CHADS₂ score [16]. All data analyzes were performed using SAS statistical software (Version 9.4, SAS Institute, Cary, NC). The two-sided significance level was set at p < .05.

Results

A total of 16,888 PAOD patients was included; among them, 76.8% patients were aged ≦74 years, and 52.9% were men (Table 1). Mean age of the study patients was 63.4 (±14.5) years. The four major comorbidities of the study patients were hypertension (68.1%), hyperlipidemia (43.0%), diabetes mellitus (25.7%), and COPD (22.7%). Most PAOD patients had a CHADS₂ score of 1 (35.0%). Mean CHADS₂ score was 1.49 (±1.30). Mean follow-up period was 5.35 (±3.46) years for LEA and 5.42 (±3.43) years for mortality.

Table 1. Baseline characteristics of PAOD patients.

	PAOD
Variable	N = 16888
Age, years	n (%)
≦74	12,963 (76.8)
≧75	3925 (23.2)
Mean ± SD	63.4 (14.5)
Sex
Female	7952 (47.1)
Male	8936 (52.9)
Underlying disease(components of the CHADS2 score)
CHF	1688 (10.0)
Diabetes mellitus	4335 (25.7)
CVA or TIA	1894 (11.2)
Hypertension	11,504 (68.1)
Other underlying disease
AF	514 (3.04)
Hyperlipidemia	7257 (43.0)
COPD	3825 (22.7)
Chronic kidney disease	3191 (18.9)
Gout	3061 (18.1)
CHADS2 score
0	3961 (23.5)
1	5905 (35.0)
2	3898 (23.1)
3	1618 (9.58)
4	950 (5.63)
5	451 (2.67)
6	105 (0.62)
Mean CHADS2 score (SD)	1.49 (1.30)
Mean follow-up of amputation, y (SD)	5.35 (3.46)
Mean follow-up of mortality, y (SD)	5.42 (3.43)

PAOD, peripheral arterial occlusive disease; SD, standard deviation; CHF, congestive heart failure; CVA, cerebral vascular accident; TIA, transient ischemic attack; AF, atrial fibrillation; COPD, chronic obstructive pulmonary disease.

The incidence of LEA increased from 0.45% to 9.36% while the CHADS₂ score increased from 0 to ≧4 (Figure 1(A)). The incidence of death increased from 3.66% in PAOD patients with CHADS₂ score 0 to 31.67% in those with CHADS₂ score ≧4 (Figure 1(B)).

Figure 1. The incidence of lower extremity amputation (A) and death (B) according to CHADS₂ score in patients with peripheral arterial occlusive disease.

After adjusting for AF, hyperlipidemia, COPD, CKD, and gout, as compared with PAOD patients with a CHADS₂ score of 0, the corresponding adjusted HR (aHR) for incident LEA were 2.37 (95% CI = 1.40–4.01), 7.78 (95% CI = 4.64–13.0), 14.2 (95% CI = 8.24–24.3), and 20.8 (95% CI = 12.1, 35.8) for those with a CHADS₂ score of 1, 2, 3, and ≧4, respectively (trend test, p < .001) (Table 2). As compared with PAOD patients with a CHADS₂ score of 0, the corresponding aHR for death for CHADS₂ score 1, 2, 3, and ≧4, were 2.07 (95% CI = 1.72–2.50), 4.62 (95% CI = 3.84–5.57), 7.58 (95% CI = 6.21–9.27), and 13.2 (95% CI = 10.9–16.1), respectively (trend test, p < .001).

Table 2. Incidence and HR of lower extremity amputation and death in PAOD patients stratified by CHADS₂ score.

CHADS2 score	N	No. of events	Person-years	Rate#	Crude HR	95% CI	Adjusted HR^a	95% CI
Amputation
0	3961	18	23,885	0.75	1	(Reference)	1	(Reference)
1	5905	71	35,536	2.00	2.65	(1.58, 4.44)***	2.37	(1.40, 4.01)**
2	3898	159	19,692	8.07	9.98	(6.13, 16.2)***	7.78	(4.64, 13.0)***
3	1618	110	6537	16.8	18.9	(11.5, 31.1)***	14.2	(8.24, 24.3)***
≧4	1506	141	4736	29.8	29.9	(18.3, 49.0)***	20.8	(12.1, 35.8)***
p for trend					<.001		<.001
Death
0	3961	145	23,951	6.05	1	(Reference)	1	(Reference)
1	5905	451	35,721	12.6	2.08	(1.72, 2.51)***	2.07	(1.72, 2.50)***
2	3898	590	20,107	29.3	4.97	(4.15, 5.97)***	4.62	(3.84, 5.57)***
3	1618	341	6768	50.4	8.79	(7.23, 10.7)***	7.58	(6.21, 9.27)***
≧4	1506	447	5003	89.4	15.9	(13.2, 19.2)***	13.2	(10.9, 16.1)***
p for trend					<.001		<.001

Rate#, per 1000 person-year; Crude HR, crude hazard ratio.

^a Adjusted for atrial fibrillation, hyperlipidemia, chronic obstructive pulmonary disease, chronic kidney disease, and gout.

** p < .01.

*** p < .001.

Cumulative incidence curves of LEA and death stratified by CHADS₂ score in PAOD patients are clearly demonstrated by Kaplan–Meier analyzes in Figure 2(A, B). The AUROC of CHADS₂ score in predicting LEA and death was 0.75 (95% CI = 0.73–0.77) and 0.70 (95% CI = 0.69–0.71), respectively (Figure 3(A,B)). The CHADS₂ score had an acceptable discrimination power for LEA (C-index = 0.79) and death (C-index = 0.76) based on Cox-regression analysis (Harrell C) [16].

Figure 2. Cumulative incidence curves of lower extremity amputation (A) and death (B) stratified by CHADS₂ score in patients with peripheral arterial occlusive disease.

Figure 3. Receiver operating characteristic (ROC) curve for CHADS₂ score in predicting lower extremity amputation (A) and death (B) in patients with peripheral arterial occlusive disease.

Discussion

The predictive role of CHADS₂ in incident LEA and all-cause mortality in people with PAOD has been shown for the first time in the present study.

Several predictors of lower limb amputation and mortality among PAOD patients have been defined [17–19]. However, limited scoring systems were proposed for predicting extremity amputation and mortality in this specific patient population [20–25]. Moreover, the proposed scoring systems are relative impractical for clinical use due to their complicatedness [20–25]. Hence, the current study aimed to clarify whether the CHADS₂ score, a ready-made product capturing several well recognized atherothrombotic risk factors [4–8], served as an optimized risk scoring scheme for limb loss and mortality among people with PAOD. The presented results were comparable with prior studies [20–25]; suggesting that CHADS₂ score might be considered as a viable alternative of risk stratification tool in this population.

Early study has shown that an elevated CHADS₂ score correlates with the end point of critical limb ischemia in people with PAOD [26]. However, it should be noted that no attempt was made to assess the discriminative ability in that study since the odds ratio but not the AUROC or Cox C-index analysis was implemented. Moreover, the study conducted by Gary et al. was limited by the single-center design with relative small sample size. Furthermore, the hard outcome, such as mortality, was not included for analysis. In contrast, the main strength of our study is its large-scale, 16,888 study patients with an extended follow-up period. Our findings further diversified the predictive role of the CHADS₂ score in risk stratification for LEA and mortality among PAOD patients. Further prospective investigations might be needed to verify the presented findings.

In this study, we found that people with PAOD were at an increased risk of lower limbs amputation and death while their CHADS₂ scores were 1 or above, as compared with those with a score of 0. In addition, the CHADS₂ score was reported to confer an acceptable power to predict the outcomes in PAOD patients. Based upon its inherent nature of convenience and simplicity, CHADS₂ score might be helpful for the physicians for risk discrimination of LEA and mortality in this population.

Limitations

First, personal health behaviors were not available in this dataset. Furthermore, the information on the severity of PAOD, such as ankle-brachial index, Fontaine classification and Rutherford classification, and on therapies for PAOD, such as medical treatment, endovascular stenting and surgical intervention were not included in this study. Second, the diagnoses were made by ICD coding system. Finally, comparison between using the CHADS₂ scoring system as a predictor of the outcomes and other outcomes-prediction tools in this patient population cannot be performed due to the limitations of this database.

Conclusions

This study reported for the first time that CHADS₂ score was predictive of the outcomes of LEA and all-cause mortality in PAOD patients. Acceptable predictive ability was noted; hence, the CHADS₂ score might be useful for the physicians while caring for these patients for risk stratification.

Disclosure statement

No potential conflict of interest was reported by the authors..

Additional information

Funding

This study was supported in part by Taiwan Ministry of Health and Welfare Clinical Trial and Research Center of Excellence [MOHW106-TDU-B-212-113004], China Medical University Hospital, Academia Sinica Taiwan Biobank Stroke Biosignature Project [BM10601010036], Taiwan Clinical Trial Consortium for Stroke [MOST 106-2321-B-039-005], Tseng-Lien Lin Foundation, Taichung, Taiwan, Taiwan Brain Disease Foundation, Taipei, Taiwan, and Katsuzo and Kiyo Aoshima Memorial Funds, Japan.