Is Heart Failure More Prevalent in Patients With Peripheral Arterial Disease? A Meta-Analysis

Authors

  • Rishi G. Anand MD,

    1. From the Department of Cardiovascular Medicine, Ochsner Clinic Foundation, New Orleans, LA;1 and the Division of Cardiology, University of Maryland School of Medicine, Baltimore, MD2
    Search for more papers by this author
  • 1 Hector O. Ventura MD,

    1. From the Department of Cardiovascular Medicine, Ochsner Clinic Foundation, New Orleans, LA;1 and the Division of Cardiology, University of Maryland School of Medicine, Baltimore, MD2
    Search for more papers by this author
  • and 1 Mandeep R. Mehra MD 2

    1. From the Department of Cardiovascular Medicine, Ochsner Clinic Foundation, New Orleans, LA;1 and the Division of Cardiology, University of Maryland School of Medicine, Baltimore, MD2
    Search for more papers by this author

Rishi Anand, MD, Department of Cardiology, Ochsner Clinic Foundation, 1514 Jefferson Highway, New Orleans, LA 70121
E-mail: ranand@ochsner.org

Abstract

Because risk factors for heart failure (HF) cluster in persons with peripheral artery disease (PAD), the authors conducted a meta-analysis to examine the prevalence of HF in individuals with PAD. MEDLINE searches were performed to review all PAD clinical trials (1966–2003). Expected control population prevalence rates for HF were derived from the National Health and Nutrition Examination Survey (NHANES) database. In total, 11,304 patients were evaluated. The average age of the patients was 67±5 years. The prevalence of HF in patients with PAD was 7.9% (range, 5.3%–13.9%) compared with an expected prevalence of 4.1%(range, 3.7–4.5%). The relative risk for increased HF prevalence among those with PAD was 1.9 (range, 1.35–3.10; P<.001). Thirteen (range, 7–19) PAD patients needed to be screened to detect 1 case of HF. The presence of PAD is associated with a 2-fold increase in the prevalence of HF. The use of PAD as a risk marker for underlying HF may enhance the effectiveness of screening criteria for HF detection.

The development of heart failure (HF) represents a late event in the cardiovascular continuum and is largely the result of the concerted influence of risk factors that mediate atherosclerosis. Risk factors for HF (ie, atherosclerosis, smoking, hypertension, and diabetes) cluster in patients with peripheral artery disease (PAD). Worldwide, the most common cause of PAD is atherosclerosis.1 PAD has been established as an independent risk factor for mortality among patients with coronary artery disease (CAD), as well as a risk factor for predicting prevalence of CAD among patients.2 We go one step further by hypothesizing that patients with PAD have a higher prevalence of HF than the national age- and sex-matched prevalence rates for HF.

The purpose of this study was to perform a systematic review using meta-analysis to ascertain the prevalence of HF among patients with PAD, a hypothesis that has not been previously investigated.

Methods

Search Strategy. A detailed computerized search in MEDLINE (1966–2003) was performed to find all studies that addressed PAD of the lower extremities. The following keywords were used: peripheral vascular disease, PAD, peripheral arterial occlusive disease, ankle-brachial index (ABI), claudication, intermittent claudication, and claudicant. The search was limited to published trials with human participants who spoke English and with a patient enrollment age older than 19 years. Reference citations from included studies were also reviewed. Personal communications with recognized experts in HF and PAD were performed to identify studies that escaped our search parameters.

Inclusion Criteria. All of the following criteria had to be fulfilled for study inclusion: (1) Prospective trial on PAD or its medical/surgical management. A patient was considered to have PAD if he or she had an ABI ≤0.90, documentation of PAD or intermittent claudication in the medical record, femoral bruits, or nonpalpable pedal pulses or was undergoing lower extremity vascular surgery for management of PAD. (2) Study population >100 patients.(3) Baseline characteristics of studies had to delineate prevalence of HF among its study population.

Exclusion Criteria. All of the following criteria led to exclusion from our review: (1) Study population <100 patients. (2) Restrictive populations not reflective of PAD populations on the whole. A study evaluating patients with PAD and meningitis would be considered a restrictive population. (3) Restrictive intervention study populations not reflective of PAD populations on the whole. A study evaluating patients with PAD undergoing coronary angioplasty would be considered a restrictive intervention. (4) Study baseline characteristics matching. (5) Study includes supraumbilical arterial (ie, abdominal or cerebrovascular) disease in its definition of PAD. (6) Study includes acute arterial thrombosis as a cause of PAD. (7) Retrospective study.(8) Redundant study.

Our search method identified approximately 500 studies. After applying exclusion/inclusion criteria, 7 studies were determined eligible for meta-analysis.

Data Abstraction. All original articles were retrieved for data abstraction and evaluated by 2 reviewers. Disagreements regarding study eligibility were resolved by consensus. If no consensus was reached, the opinion of an independent reviewer was decisive. Study period, publish date, study design, ABI, mean age, total number of patients with PAD, HF prevalence among study patients, geographic area of study conduction, and patient population demographics were recorded on a standard form.

Expected prevalence rates of HF to serve as population controls were derived from the age- and sex-specific National Health and Nutrition Examination Survey (NHANES) epidemiologic database.3 NHANES was chosen over other studies, notably the Framingham Heart Study, because the study was designed to determine the prevalence of HF in nonhospitalized men and women in the United States. NHANES defined HF in 2 manners. First, participants were asked on a self-report questionnaire if they had ever been told by a physician that they had HF. An operative definition of HF was also devised based on criteria from the Framingham Heart Study and the World Health Organization dyspnea questionnaire.4 HF was defined as a total clinical score of >3 (Table I). We utilized clinical score-derived expected prevalence rates for HF when performing the meta-analysis (Table II).

Table I.  Heart Failure Clinical Score
Clinical VariablesScore
Dyspnea/difficulty breathing 
 Trouble with breathing (shortness of breath) 
  Hurrying on the level or up slight hill1
  At ordinary pace on the level1
  Do you stop for breath when walking at own paceinline image2
  Do you stop for breath after 100 yards on the levelinline image2
Physical examination 
 Heart rate, bpm 
  91–1101
  ≥1112
 Rales/crackles 
  Either lower lung field1
  Either lower and either upper lung field2
 Jugulovenous distention 
  Alone1
  Plus edema2
  Plus hepatomegaly2
 Chest radiography 
  Cephalization of pulmonary vessels1
  Interstitial edema2
  Alveolar fluid plus pleural fluid3
  Interstitial edema plus pleural fluid3
Abbreviation: bpm, beats per minute. Adapted from Schocken et al.3
Table II.  Estimated HF Prevalence as Assessed by Clinical Score: Noninstitutionalized US Population Aged 25–74 Years, NHANES
Sex and Age, yPrevalence, %Estimated No. of Cases
Women
 25–541.3500,990
 55–643.0303,137
 65–744.3316,816
 Total 25–742.01,120,988
Men
 25–540.8295,477
 55–644.5407,306
 65–744.8273,713
 Total 25–741.9976,496
Women and men
 25–541.1796,467
 55–643.7710,443
 65–744.5590,574
 Total 25–742.02,097,484
Abbreviations: HF, heart failure; NHANES, National Health and Nutrition Examination Survey. Adapted from Schocken et al.3

Statistical Analyses. For purposes of the meta-analysis, the pooled data were analyzed using Mantel-Haenszel testing in a random effects model that included tests for heterogeneity.

Results

Patients. The 7 studies (Table III) that met inclusion criteria included 10,205 patients for analysis. The entire study population had a mean age of 67 years and a mean ABI of ≤0.90. Only one study5 delineated prevalence of HF with respect to sex. One study had an international patient enrollment. Five studies were performed in the United States and one study was performed in France. Because of limited study size, no significant relationship between ABI and HF prevalence could be ascertained.

Table III.  Baseline Characteristics of Studies Analyzed
Source (Publication Date)Study PeriodDesignNo.Mean Age, yABICHF Prevalence, %Expected Prevalence Clinical Score, %
Hirsch (2001)131999OBS82571<0.906.064.50
Newman (1999)141989–1990OBS76874<0.9013.94.50
Becquemin (1997)151989–1992RCT243670.449.054.50
McDermott (1997)161995RCT20268<0.908.004.50
CAPRIE Steering Committee (1996)171992–1995RCT645264<0.856.003.70
Vogt (1993)51977–1989RCT145270<0.9010.054.50
Wilson (1989)181983–1989RCT263620.5055.833.70
Abbreviations: ABI, ankle brachial index; CHF, congestive heart failure; OBS, prospective observational; RCT, randomized control trial.

Pooled Analysis. The Figure depicts the results of the meta-analysis. All studies had a higher prevalence of HF than age- and sex-matched expected prevalence rates for HF as determined by NHANES. The mean prevalence of HF in patients with PAD was 7.3%(range, 5.3%–13.9%) compared with an expected population prevalence based on clinical score of 4.1% (range, 3.7%–4.5%). The relative risk for an increased HF prevalence among those with PAD was 1.9 (95% confidence interval, 1.68–1.94; P<.001). The number needed to screen PAD patients to detect 1 case of HF was 32 (95% confidence interval, 28–38).

Figure Figure.

Results of the meta-analysis, depicting a significantly increased relative risk for increased heart failure prevalence in the presence of peripheral artery disease.

Discussion

Traditional risk factors for PAD have been well defined. These risk factors include smoking, increasing age, diabetes, hypertension, hyperlipidemia, and low high-density lipoprotein cholesterol levels.6 Moreover, these are the exact same risk factors as CAD. Numerous studies have proven that atherosclerotic disease in one vascular bed predicts atherosclerotic disease in another vascular bed. Mautner and colleagues7 performed a necropsy study analyzing the prevalence of CAD in patients who underwent amputation of 1 or both lower extremities because of severe PAD. They found that 92% of the patients studied had 76% to 100% narrowing in a cross-sectional area of 1 or more major coronary arteries by atherosclerotic plaque.7 Similarly, Aronow and Ahn8 reported on the prevalence of coexistence of CAD and PAD in more than 1600 patients living in a long-term health care facility. They found that if PAD was present, CAD was also present in 58% of the population. The coexistence of CAD and PAD has even been documented in patients with vasculogenic erectile dysfunction where 19% of patients with erectile dysfunction had clinically silent CAD documented by angiography.9 With the high prevalence of PAD and CAD cohabitation firmly established and the fact that CAD is a substantial cause of HF in the Western world,10 evaluating the prevalence rates of HF in patients with PAD seems worthwhile.

The results of this meta-analysis confirm the hypothesis that patients with PAD have a higher prevalence of HF than the national age- and sex-matched prevalence rates for HF. Furthermore, PAD status may serve as an important factor in the development of a screening strategy for HF. Based on the results of our study, 32 patients with PAD needed to be screened to detect 1 case of clinically apparent HF. In comparison, the number needed to screen using mammography to detect 1 case of breast cancer is 1792 (for women older than 50 years).11 Therefore, using PAD as a risk marker to enhance HF screening appears to be worthwhile when compared with a nationally accepted screening tool.

A cost-effective screening strategy that adequately identifies patients with clinically silent ventricular dysfunction has yet to be developed, however. A recent study by Heidenreich and associates12 suggests that B-type natriuretic peptide testing followed by echocardiography is likely to be a cost-effective screening strategy for men aged 60 years and possibly for women when at least a 1% prevalence of moderate or greater left ventricular systolic dysfunction exists. The cost effectiveness increases even further with populations where the estimated prevalence of heart failure is even greater than 1%, such as in PAD patients.

Based on our review, it may be worthwhile to use PAD as part of a screening protocol to identify HF patients. One possible method could focus on having all PAD patients and their physicians complete a survey, such as in Table I. A certain numeric score greater than a predetermined threshold would trigger further screening tests such as B-type natriuretic peptide assays or possibly even echocardiography.

Furthermore, the studies included for analysis delineated HF with respect to clinically apparent systolic dysfunction. Patients with diastolic HF or those with structural heart changes but no symptoms were not included in the review. Unfortunately, a rigorous national estimation of the scope of diastolic dysfunction remains to be performed. Therefore, if diastolic HF had been reported, we estimate that PAD would be an even stronger predictor of HF.

Conclusions

The presence of PAD is associated with a nearly 2-fold increase in the prevalence of HF. It is our contention that strategies that use PAD as a potent risk marker for underlying HF might serve to further enhance the cost effectiveness of general screening criteria for the underlying detection of HF.

Ancillary