Outcome events in patients with claudication: A 15-year study in 2777 patients☆☆☆★

Article Outline

• Abstract
• Methods
• Results
• Discussion
• Conclusions
• Discussion
• References
• Copyright

Abstract 

Objective: The purpose of this study was to delineate the natural history of claudication and determine risk factors for death. Methods: We reviewed the key outcomes (death, revascularization, amputation) in 2777 male patients with claudication identified over 15 years at a Veterans Administration hospital with both clinical and noninvasive criteria. Patients with rest pain or ulcers were excluded. Data were analyzed with life-table and Cox hazard models. Results: The mean follow-up was 47 months. The cohort exhibited a mortality rate of 12% per year, which was significantly (P < .05) more than the age-adjusted US male population. Among the deaths in which the cause was known, 66% were due to heart disease. We examined several baseline risk factors in a multivariate Cox model. Four were significant (P < .01) independent predictors of death: older age (relative risk [RR] = 1.3 per decade), lower ankle-brachial index (RR = 1.2 for 0.2 change), diabetes requiring medication (RR = 1.4), and stroke (RR = 1.4). The model can be used to estimate the mortality rate for specific patients. Surprisingly, a history of angina and myocardial infarction was not a significant predictor. Major and minor amputations had a 10-year cumulative rate less than 10%. Revascularization procedures occurred with a 10-year cumulative rate of 18%. Conclusions: We found a high mortality rate in this large cohort and four independent risk factors that have a large impact on survival. Risk stratification with our model may be useful in determining an overall therapeutic plan for claudicants. A history of angina and myocardial infarction was not a useful predictor of death, suggesting that many patients in our cohort presented with claudication before having coronary artery symptoms. Our data also indicate that claudicants have a low risk of major amputation at 10-year follow-up. (J Vasc Surg 2001;33:251-8.)

Intermittent claudication is a common clinical condition estimated to affect at least 10% of individuals older than 70 years.¹, ², ³ It is widely agreed that adverse limb outcomes such as gangrene and amputation are relatively rare among patients with claudication. However, adverse outcomes of systemic atherosclerosis, including death, are common.³ Even after adjustment for known risk factors, claudicants exhibit a higher mortality rate than the healthy population.³ Although these generalizations are undoubtedly true in the aggregate, there is a paucity of data to help the clinician predict outcomes in an individual patient. Relatively few studies of outcome prediction have focused specifically on claudicants.⁴, ⁵, ⁶, ⁷, ⁸ Some of these reports⁷, ⁸ predate the widespread use of noninvasive vascular testing, whereas others are small or have short follow-up.⁴, ⁵, ⁶

In this current study, we report an analysis of long-term outcomes among 2777 male claudicants identified with both clinical history and noninvasive laboratory criteria. The outcomes include limb revascularization, amputation, and death. We focus in particular on the risk factors that help to predict the mortality rate among these patients.

Back to Article Outline

Methods 

This prospective study period extended from January 1983 to January 1998. During this time more than 28,000 lower extremity vascular laboratory studies, representing 4669 patients, were performed at the Pittsburgh Veterans Administration (VA) Medical Center. At each visit to the laboratory, a registered nurse obtained a detailed lower extremity vascular history of each patient. Most patients were considered to be reliable, and clinical variables were obtained through patient interviews. A history of smoking was obtained by questioning the patient. Racial information was reported by the patients. Patients were questioned about whether the following conditions had been diagnosed by their physicians: hypertension, diabetes, angina, myocardial infarction (MI), and stroke. In the case of a positive answer, a yes was recorded for angina, MI, and stroke. For diabetes and hypertension, the patient was asked about the type of therapy, and the answer was recorded. A yes answer for angina, stroke, and MI was confirmed with further questioning about specific symptoms to ensure that the patient correctly understood the question. In this context, angina was defined as exertional chest pain or discomfort alleviated by rest or nitroglycerin, and stroke was defined as a focal neurologic deficit lasting more than 24 hours. For the small proportion of patients who were unreliable (estimated at < 5% of the total), medical history information was obtained from the computerized medical record system. Nurses had the option of leaving any field blank if a definite affirmative or negative answer could not be obtained. Blood pressures were obtained in both arms, and a lower extremity study was performed. For this study, we used the baseline values of each of the analyzed variables (ie, the value recorded at the time of the first study). Females were excluded from the current study because few women presented for vascular laboratory testing at the Pittsburgh VA during the study period. Male patients were identified as claudicants if they met all of the following criteria:

1.Resting ankle-brachial index (ABI) on the symptomatic side less than 0.9, or

2.Treadmill exercise–induced reduction of ABI by more than 0.29, or

3.Among patients with noncompressible vessels (ABI > 1.25), elimination of reflected wave and a more than 50% reduction in amplitude of pulse volume recording (PVR) at ankle, in comparison with thigh PVR

Among patients meeting criteria 1, 2, and 3, we identified 2418 patients meeting criterion 4, a. An additional 194 patients met criterion 4, b, and a final group of 165 patients met criterion 4, c. This yielded the study group of 2777 patients.

To estimate the validity of the medical history information collected by the nurses in the vascular laboratory, one of the authors performed a blinded review of the medical records of 241 randomly selected patients. In this manner we obtained independent medical record information about hypertension, diabetes, angina, MI, cerebrovascular accident (CVA), and smoking history. Comparison of these data with the vascular laboratory data for these variables showed an excellent agreement for each variable: hypertension, 88%; diabetes, 91%; angina, 92%; MI, 94%; CVA, 90%; and smoking, 93%.

Outcome events were extracted from two VA national databases (beneficiary identification and record locator subsystem and the national patient care database) as well as from hospital records. The national VA databases have been validated with studies that show accuracy rates of more than 95% by comparison with direct patient record review (described at www.virec.research.med.va.gov ). The follow-up databases we used for this study are designed to track veteran mortality rates regardless of the location of death. In addition, revascularization and amputation procedures are tracked for all VA hospitals. This allowed us to track these outcomes wherever they may have occurred in the VA system. However, revascularization and amputation may be incomplete because of the possibility of veterans undergoing some procedures outside the VA. Missing data across the database were relatively infrequent, ranging from 0% for the most fully populated field (patient identification) to 8% for the field with the greatest proportion of missing data (pulse pressure).

Point values for different risk factors in multivariate modeling were developed with previously described methods.⁹ Expected survival for the cohort (the “normal” survival curve in Fig 1) was computed by the use of Hakulinen's method¹⁰ with internal rate tables in S-PLUS (Mathsoft, Cambridge, Mass).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1. 

  Kaplan-Meier survival curves for study cohort (solid line ) and for age-adjusted healthy US male population (dotted line ). A 95% CI is shown for study cohort survival curve.

With this method, a cohort survival curve is computed that is an average of the individual expected survival curves for referents from the national population, matched for age, sex, and year of entry. In addition, this method also recreates a similar censoring pattern in the expected survival function. In our analyses, we used Access (Microsoft, Redmond, Wash) for descriptive statistics and SAS (SAS Institute, Cary, NC) for Cox proportional hazards modeling, Kaplan-Meier plots, and parametric modeling.

Back to Article Outline

Results 

Table I shows general demographic and clinical characteristics of the patient cohort. Aside from the fact that all the patients are males, the cohort has characteristics typical of patients with vascular disease in other studies, including a similar age, race, smoking history, and other similar cardiovascular risk factors. The mean patient follow-up in our study was 47 months. A total of 1363 patients died during follow-up. The survival curve (Kaplan-Meier method) for the entire group is shown in Fig 1 (solid line ), along with a 95% CI. By fitting an exponential parametric model, we calculated the annual mortality risk among our patients as 12% of the at-risk group. The usefulness of this summary statistic was assessed by determining that the survival data closely fit the exponential model (residual deviance χ² < 0.0005). The mortality rate among the claudication cohort was significantly (P < .05) greater than the mortality rate of the age-adjusted “normal” male population of the United States (Fig 1, dotted line ). We were able to ascertain the cause of death for a subset consisting of 236 patients who died in the hospital. Sixty-six percent of these patients died of ischemic cardiac events.

Table I. Clinical and demographic caharcteristics

Fig 2 shows the Kaplan-Meier plot of the occurrence of three important limb-related events: minor amputation (below ankle level), major amputation (above ankle level), and surgical revascularization.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 2. 

  Kaplan-Meier curves showing probability of freedom from selected limb-related events: minor amputation, major amputation, and surgical revascularization.

Minor and major amputations occurred with a 10-year cumulative frequency of less than 10%. The 10-year cumulative frequency of surgical revascularization was 18%.

We next sought to delineate the factors that predict the time to death among our patient cohort. Table II shows the results of univariate Cox proportional hazards modeling of a number of potential predictors. The five factors that had significant predictive value were age, ABI, diabetes, history of CVA, and pulse pressure. We noted that diet-controlled diabetes was not a significant risk factor, but that diabetes that required oral agents for treatment and diabetes requiring insulin were both significant predictors with roughly equal predictive weight. Therefore, for subsequent analysis, we used the variable “diabetes requiring medication.” Using the multivariate Cox modeling of the five factors identified with univariate analysis, we found that four were independent predictors of death: age, history of CVA, ABI, and diabetes requiring medication (pulse pressure was the one variable that did not retain predictive value in the multivariate model). The upper section of Table III shows the model parameters for this initial multivariate model.

Table II. Univariate prediction of time to death in claudicants

β Values and RR were calculated with the univariate proportional hazards model. The β value is the natural logarithm of RR. In the case of continuous variables (age, smoking pack/years, ABI, pulse pressure, systolic pressure), the RR specifies the risk multiplier for a one-unit change in the variable. As a result, the RR may seem low even when the variable in question is a strong predictor. In the case of the other variables, the RR indicates the increased RR associated with having the risk factor. “Symptomatic CAD” is defined as the presence of either angina or prior MI.

ABI, Ankle-brachial index; CAD, coronary artery disease; CVA, cerebrovascular accident; MI, myocardial infarction; NS, not statistically significant; RR, relative risk.

Table III. Multivariate prediction of risk of death in claudication patients

β Values and RR were calculated with the multivariate proportional hazards model. Further description of the β and RR values can be found in Table II.

ABI, Ankle-brachial index; CVA, cerebrovascular accident; NS, not statistically significant; RR, relative risk.

A second multivariate model (Table III, lower section ) was developed after dividing age and ABI into appropriate categories. This second model lends itself to a point system, wherein any given patient can be assigned a “risk score” by adding the appropriate point values shown in Table III. Fig 3 shows the predicted 2-year mortality rate (and 95% CI) for claudicants as a function of the risk score, on the basis of our multivariate Cox model.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 3. 

  Predicted 2-year mortality rate of male claudicants as a function of risk score (computed with point system shown in Table III). A 95% CI is shown for mortality estimate. Figure also shows raw 2-year mortality rate for our study cohort, as a function of risk score (closed circles ). Number of patients with risk scores of 12 and 13 is small, so raw 2-year mortality data points for these scores should be interpreted with caution.

For comparison, we have shown the actual 2-year mortality data for our patient cohort as a function of risk score (closed circles in Fig 3).

Back to Article Outline

Discussion 

This study represents one of the largest in which the outcomes among claudicants are examined. We have documented various outcomes (mortality rate, major and minor amputation, revascularization) among a large cohort of claudicants, using methods that allow for long-term national follow-up. In comparison with older large studies of claudicants,⁷, ⁸ one strength of the current analysis is that claudicants were identified with both clinical and noninvasive criteria; thus, we avoided including patients with nonvascular leg pain. Our choice of the resting and treadmill ABI criteria (“Methods” section) for diagnosing peripheral vascular disease is consistent with several other studies.³, ¹¹, ¹², ¹³ Although practitioners routinely rely on PVR tracings when patients have noncompressible vasculature,³ there are no widely accepted criteria for what constitutes a significant reduction in PVR waveform. We used the same criterion that we use in our clinical practice (> 50% reduction of amplitude from thigh to ankle and loss of reflected wave at the ankle level). In any event, less than 6% of the study group was entered on the basis of PVR criteria.

The mortality rates observed in our study population are striking, with an annualized mortality risk of 12%. Five- and 10-year mortality rates were 42% and 65%, respectively. In comparison, Dormandy et al¹⁴ recently analyzed a large number of previous studies and found that 5- and 10-year mortality rates among claudicants averaged 30% and 50%, respectively. These differences may be related to the fact that most previous studies have included both men and women (women are lacking in our study and are known to have a longer life expectancy).

Our data further show that adverse limb ischemic events such as minor and major amputation are relatively infrequent among claudicants (Fig 2). These findings are consistent with those of many previous studies.³

We have also developed a predictive model to help stratify claudicants according to the risk of death. It is widely accepted that claudication is a marker for systemic atherosclerosis, leading to future cardiovascular death and morbidity. However, there are no well-established criteria allowing the clinician to predict outcomes in individual patients. Understanding the natural history, especially the likelihood of death, is crucial to clinical decisions, such as whether to offer surgical revascularization.

The four variables that have independent predictive value for the time to death were age, prior CVA, ABI, and diabetes requiring medication. In principle, Model 1 in Table III could be used to calculate the mortality risk for individual claudicants. However, Model 2 lends itself to a convenient point system. For any given claudicants, the appropriate point values are summed and yield a risk score. Fig 3 shows the predicted 2-year mortality rate as a function of risk score. The choice of a 2-year mortality rate was somewhat arbitrary, because we could have chosen any mortality end point for this figure. We think that a 2-year mortality prediction may be useful in patient education and in the determination of the focus of patient care. Because limb amputation is infrequent among claudicants, the primary purpose of revascularization is to improve the quality of life. Therefore, patients with a high predicted mortality rate may be better served with aggressive cardiac risk factor intervention than with aggressive intervention for claudication. In any specific case, the clinician and patient will still need to factor in the severity of the claudication and the degree of lifestyle impairment.

Our finding of the importance of an ABI as a predictor of death is consistent with previous work⁵, ¹¹, ¹⁵ and appears to confirm again the correlation between the severity of peripheral and coronary arterial disease. The important role of diabetes has also been previously demonstrated.⁵, ⁶, ¹⁶ We found that diet-controlled diabetes was not a significant predictor; however, diabetes that required either oral or insulin therapy was a strong predictor of reduced life expectancy. This finding is consistent with recent work showing that glucose control, not the mere presence of diabetes, is the critical factor in determining the likelihood of complications from diabetes. In our study group, a history of CVA was an important predictor of time to death. To our knowledge, the relevance of this variable has not been previously emphasized.

Our analysis is also notable for the variables that were found not to be significant predictors of time to death. Although smoking, angina, a history of MI, and a history of hypertension all correlated with reduced life expectancy, the correlations were not statistically significant (Table II). In the case of smoking, the issue may simply be that large numbers of patients in our study group (94%) were current or past smokers. The small number of nonsmokers may have made it difficult to determine the true predictive value of smoking. The lack of predictive value for a baseline history of angina and MI is at odds with the findings of other studies.⁵, ⁷, ¹⁷ It may be that many of our patients presented with claudication before presenting with coronary symptoms. Other means of cardiac assessment (eg, electrocardiography, nuclear imaging, cardiac catheterization) may have proved to be better predictors of death. It is also possible that patients who have a diagnosis of coronary disease are more closely followed up and treated for the disease. In any case, our data demonstrate the potential pitfall of relying on cardiac history alone as a means of predicting the mortality rate. Hypertension has been previously identified as a predictor of death among claudicants.⁵, ⁶, ⁷ Although we did not find that a history of hypertension was a significant predictor, an elevated pulse pressure was a predictor in univariate analysis (Table II). In previous work in which the progression of carotid disease was studied,¹⁸, ¹⁹ we documented the importance of pulse pressure. In the current analysis, pulse pressure did not retain significance in the multivariate models.

An important limitation of our study population is the lack of women. This is virtually inevitable, given the setting of this study (VA Hospital). Additionally, our study included few Hispanic patients, because of the makeup of the veteran population in the western Pennsylvania region. In other respects, however, the demographic and clinical characteristics of our study population (Table I) are not much different from that seen in other vascular laboratories. Female claudicants are likely to have a natural history different from males. In the interpretation of our natural history data, it should also be noted that not all claudicants are referred for noninvasive evaluation. Therefore, our results should only be extrapolated to other male claudicants with claudication of sufficient severity to be referred to a noninvasive vascular laboratory.

Although our model provides a close fit to our patient cohort (Fig 3), the true test is the applicability of the model to other male claudicants. We are currently gathering data on new patients both within and without the VA system to conduct such analysis in the future.

Back to Article Outline

Conclusions 

Among a large, prospectively followed cohort of male claudicants with long-term national follow-up, we have defined risk factors that predict death. By the use of a point system, four easily identifiable factors (age, diabetes, history of CVA, and ABI) can be used to stratify patients into groups with differing predicted mortality rates. These data may be helpful in clinical decision making in the care of claudicants.

Back to Article Outline

Discussion 

Dr Walter J. McCarthy III (Chicago, Ill). Dr Porter, Dr Riles, Dr Pearce, members, and guests. It's an honor to discuss this interesting paper.

The authors have brought to our attention the problem of intermittent claudication, which is a remarkably prevalent condition affecting as many as 10% of North Americans older than 70 years. By some estimates, there are at least 4 million to 5 million Americans with intermittent claudication, and the proper management of this huge patient cohort remains controversial. For this reason, natural history information, such as that contained in the present study, is invaluable. The authors have captured a huge group of patients within their own VA Hospital and then used two VA national databases for the follow-up end points. This is an ingenious technique because it avoids the almost impossible task of following up on all these patients using the usual methods.

My first question relates to the accuracy of these two VA databases. Are you sure that the data are really sound? Did you do some sampling of, say, 100 of your patients and then go back and use record reviews and telephone interviews to compare your own results with the databases? Did you use the national death index registry to confirm the mortality rate?

My remaining questions relate to the analysis and the presentation of the data because you certainly have a great deal of very valuable data. Which of these patients had iliac as compared with superficial femoral disease, and how did that affect their prognosis? That's a question that all of us wonder about.

Was revascularization used for limb salvage as the disease progressed over the years or was it used simply to relieve the claudication symptoms?

Did you calculate a correlation coefficient for the relationship between the ankle-brachial index and the length of life? This is a number that would be very interesting, and this relationship is something unique in your research, I think.

Finally, you couldn't show that smoking was a predictor of early death in your univariate analysis because 94% of your patients smoked; however, you could perform a subgroup analysis comparing the current smokers with the past smokers and then try and show a difference in length of life and also trying to predict the amputation rate. That would help us with our argument to patients that they should stop smoking.

I congratulate the authors on their manuscript and on their effort and look forward to seeing the final published paper. Thank you.

Dr Satish C. Muluk. Thank you, Dr McCarthy, for your comments.

An important issue that you raise is the accuracy of the databases that we used. Obviously that's central to the validity of the study. The databases have been validated by independent study, and they reportedly have an accuracy rate of greater than 95%. We did not perform any validation on our own, although we did to the extent that we could confirm deaths within the hospital. Obviously, that represents only a small subset, and we didn't think that it was valid to use that information as a way of validating the database itself. On the other hand, the data that we had available to us showed excellent validity of the VA database. But I think the key point is that there has been independent validation of these VA databases.

I would point out that one of them, the BIRLS, is actually used for benefit payments to surviving relatives, and so there is a great deal of effort that goes into making it accurate. Obviously it has to do with payments given to people who survive, and therefore, it's quite important for the government to maintain accuracy of that.

The question you asked about iliac versus SFA disease is very important to us as well. We did not study that for this paper, but that's in fact the next step that we propose to do.

One of the other questions you asked was whether the surgeries were done for claudication or limb salvage. Although we did not report it in the manuscript specifically, I can tell you that the vast majority of operations was done for claudication. Only a few of the operations, representing the small number done after the first year or two after enrollment, were done for limb salvage, representing the fact that very few of these patients actually went on to have limb-threatening ischemia.

We did correlate ankle-brachial index with the length of life. The specific correlation there is not evident in the fairly sophisticated statistical analysis we did because we felt it was more useful to view it as a part of a multivariate model, but there is no question that independently, as a single factor, ankle-brachial index does predict length of life in a statistically striking manner. And the risk ratio associated with that univariate analysis is about 1.4 for every 0.2 points of change in the ankle-brachial index.

Finally, the issue of smoking you raised, whether we compared current versus past smoking as a factor. We did indeed do this analysis, but found that it was not an important factor. I honestly think that a lot of it has to do with patients simply not being truthful about their smoking habits. And probably the percentage of current smokers is truly higher than we indicate because patients know what to tell their doctors to make them happy, and I think that's what's going on in this instance.

Dr Donald L. Jacobs (St Louis, Mo). Your patient population is a vascular lab patient population. Since that's a different population than the one that would actually get vascular surgery consultation, do you know the percentage of these patients that were referred to the vascular clinic?

Dr Muluk. That's a good question. I'm afraid I don't have that answer. I would suspect, because of the structure of our VA, that a large number of them would have been seen in the vascular clinic because we typically see all of the claudicants at least once; it's just the nature of our setup at the Pittsburgh VA. So I would think the percentage is quite high, but I don't know the exact number.

Dr Porter. But you do know that vascular lab–selected patients have already gone through a preselection process?

Dr Muluk. They have gone through a preselection done by some physician, that's absolutely correct.

Dr Enrique Criado-Pallares (Madrid, Spain). Obviously, the ABI has a strong prognostic value for revascularization and limb salvage. You didn't tell us what the amputation and revascularization rate was for the patients with ABIs below 0.3.

Dr Muluk. Although there was a trend toward lower ankle-brachial index and major amputation, the trend was not statistically significant. And I think perhaps it's telling us that patients with low ankle-brachial index who only have claudication as a symptom are still not at that high a risk for limb loss, at least that's what our data would suggest.

Dr Keith D. Calligaro (Philadelphia, Pa). Did I understand you right that only approximately 400 patients underwent exercise studies?

Dr Muluk. No. A large number of patients underwent exercise studies, but what I showed there was that if a patient had an ankle-brachial index less than 0.9, they were enrolled as a claudicant on that basis alone. The other numbers represented the patients who had a resting ankle-brachial index greater than 0.9 but were still enrolled as claudicants because of an appropriate reduction with exercise.

Dr Calligaro. How many underwent exercise studies?

Dr Muluk. To be honest, I can't give you that number off the top of my head, but my guess is that roughly half of the patients who present for claudication undergo exercise treadmill testing.

Dr Calligaro. I would submit that potentially instead of this being a report of 2800 claudicants that it may be a report of 1400 claudicants. Although the majority of patients with a resting ankle-brachial index less than 0.9 and with classic symptoms will probably have claudication, I don't think you can assume they all have claudication on an arterial basis; they may have it because of some other cause. Your data may be somewhat flawed because you may be including some patients with lumbosacral spine disease who are having calf aching.

Dr Muluk. Well, I'd being willing to compromise and go down to 2400, but I really think that the majority of patients who have resting ABIs less than 0.9 and appropriate symptoms probably have claudication. But I do agree with you that I think in a more modern study that exercise treadmill testing would be done for everyone. This study was initiated many years ago, and of course, it wasn't as prevalent.

Dr Porter. Dr Calligaro, would you have insisted on a 20% pressure drop for study entry in all patients who presented?

Dr Calligaro. If they could complete the study.

Dr Porter. Okay. Well, I just wanted to get your position there.

Dr Robert B. Patterson (Providence, RI). At what point in their claudication history were these patients? I was interested that your curves were very flat after 2 years for amputation and revascularization, and I wonder if you can correlate that with the duration of symptoms? Were you catching some patients on a progressive downward trend and the rest of them represented a stable population?

Dr Muluk. That's an excellent question. Unfortunately, the only point that we have for certainty is the point of their first vascular lab study and that was the point of origin for this study. We did not actually record, which is an oversight, the length of symptoms prior to presentation at the lab. Therefore, I can't answer that question although it's a very good one.

Back to Article Outline

References 

1. Peabody CN, Kannel WB, McNamara PM. Intermittent claudication: surgical significance. Arch Surg. 1974;109:693–697
   • View In Article
   • MEDLINE
2. Droller H, Pemerton J. Cardiovascular disease in a random sample of elderly people. Br Heart J. 1953;15:199–201
   • View In Article
   • MEDLINE
   • CrossRef
3. TASC Working Group . Management of peripheral arterial disease (PAD). J Vasc Surg. 2000;31:S1–S289
   • View In Article
   • MEDLINE
   • CrossRef
4. Banerjee AK, Pearson J, Gilliland EL, Goss D, Lewis JD, Stirling Y, et al.  A six year prospective study of fibrinogen and other risk factors associated with mortality in stable claudicants. Thrombo Haemost. 1992;68:261–263
   • View In Article
5. Dormandy JA, Murray GD. The fate of the claudicant—a prospective study of 1969 claudicants. Eur J Vasc Surg. 1991;5:131–133
   • View In Article
   • MEDLINE
   • CrossRef
6. Jelnes R, Gaardsting O, Hougaard JK, Baekgaard N, Tonnesen KH, Schroeder T. Fate in intermittent claudication: outcome and risk factors. BMJ. 1986;293:1137–1140
   • View In Article
   • CrossRef
7. Begg TB, Richards RL. The prognosis of intermittent claudication. Scott Med J. 1962;7:342–352
   • View In Article
8. Bloor K. Natural history of arteriosclerosis of the lower extremities. Ann R Coll Surg Engl. 1961;28:36–51
   • View In Article
9. Landefeld CS, Cook EF, Flatley M, Weisberg M, Goldman L. Identification and preliminary validation of predictors of major bleeding in hospitalized patients starting anticoagulant therapy. Am J Med. 1987;82:703–713
   • View In Article
   • Abstract
   • Full-Text PDF (1267 KB)
   • CrossRef
10. Haukulinen T. Cancer survival corrected for heterogeneity in patient withdrawal. Biometrics. 1982;38:933–942
    • View In Article
    • MEDLINE
    • CrossRef
11. McDermott MM, Feinglass J, Slavensky R, Pearce WH. The ankle-brachial index as a predictor of survival in patients with peripheral vascular disease. J Gen Intern Med. 1994;9:445–449
    • View In Article
    • MEDLINE
    • CrossRef
12. Newman AB, Sutton-Tyrrell K, Vogt MT, Kuller LH. Morbidity and mortality in hypertensive adults with a low ankle/arm blood pressure index [see comments]. JAMA. 1993;270:487–489
    • View In Article
    • MEDLINE
13. Wolf EA, Sumner DS, Strandness DE. Correlation between nutritive blood flow and pressure in limbs of patients with intermittent claudication. Surg Forum. 1972;23:238–239
    • View In Article
    • MEDLINE
14. Dormandy JA, Heeck L, Vig S. Peripheral arterial occlusive disease: clinical data for decision making. Semin Vasc Surg. 1999;12:93–162
    • View In Article
15. Vogt MT, Cauley JA, Newman AB, Kuller LH, Hulley SB. Decreased ankle/arm blood pressure index and mortality in elderly women [see comments]. JAMA. 1993;270:465–469
    • View In Article
    • MEDLINE
16. Leng GC. Questionnaires. In:  Fowkes FGR editors. Epidemiology of peripheral vascular disease. London: : Springer-Verlag; 1991;p. 29–40
    • View In Article
17. Juergens JL, Barker NW, Hines EA. Arterosclerosis obliterans: review of 520 cases with special reference to pathogenic and prognostic factors. Circulation. 1960;21:188–195
    • View In Article
    • MEDLINE
18. Garvey L, Makaroun MS, Muluk VS, Webster MW, Muluk SC. Etiology of carotid stenosis progression: a 10-year study in 905 patients. J Vasc Surg. 2000;31:1–38
    • View In Article
    • Full Text
    • Full-Text PDF (97 KB)
    • CrossRef
19. Muluk SC, Muluk VS, Sugimoto H, Rhee RY, Trachtenberg J, Steed DL, et al.  Progression of asymptomatic carotid stenosis: a natural history study in 1004 patients. J Vasc Surg. 1999;29:208–214
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (100 KB)
    • CrossRef