Prediction of Mortality and Morbidity With a 6-Minute Walk Test in Patients With Left Ventricular Dysfunction

Vera Bittner, MD; Debra H. Weiner, MPH; Salim Yusuf, MBBS, DPhil; William J. Rogers, MD; Kevin M. McIntyre, MD; Shrikant I. Bangdiwala, PhD; Marvin W. Kronenberg, MD; John B. Kostis, MD; Robert M. Kohn, MD; Maureen Guillotte, RN; Barry Greenberg, MD; Patricia A. Woods, RN; Martial G. Bourassa, MD; for the SOLVD Investigators

Objective.\p=m-\Tostudy the potential usefulness of the 6-minute walk test, a self\x=req-\ paced submaximal exercise test, as a prognostic indicator in patients with left ventricular dysfunction. Design.\p=m-\Datawere collected during a prospective cohort study, the Studies of Left Ventricular Dysfunction (SOLVD) Registry Substudy. Setting.\p=m-\Twentytertiary care hospitals in the United States, Canada, and Belgium. Participants.\p=m-\Astratified random sample of 898 patients from the SOLVD Registry who had either radiological evidence of congestive heart failure and/or an ejection fraction of 0.45 or less were enrolled in the substudy and underwent a detailed clinical evaluation including a 6-minute walk test. Patients were followed up for a mean of 242 days. Outcome Measures.\p=m-\Mortalityand hospitalization. Results.\p=m-\Duringfollow-up, 52 walk-test participants (6.2%) died and 252 (30.3%) were hospitalized. Hospitalization for congestive heart failure occurred in 78 participants (9.4%), and the combined endpoint of death or hospitalization for congestive heart failure occurred in 114 walk-test participants (13.7%). Compared with the highest performance level, patients in the lowest performance level had a significantly greater chance of dying (10.23% vs 2.99%; P=.01), of being hospitalized (40.91% vs 19.90%; P=.002), and of being hospitalized for heart failure (22.16% vs 1.99%; P<.0001). In a logistic regression model, ejection fraction and distance walked were equally strong and independent predictors of mortality and heart failure hospitalization rates during follow-up. Conclusion.\p=m-\The6-minute walk test is a safe and simple clinical tool that strongly and independently predicts morbidity and mortality in patients with left ventricular dysfunction.
(JAMA. 1993;270:1702-1707)
From the Department of Medicine, University of Alabama at Birmingham (Drs Bittner and Rogers); the Veterans Affairs Medical Center, Manchester, NH (Ms and the Montreal (Quebec) Heart Institute Research Center (Dr Bourassa). Dr Yusuf is now affiliated with the Department of Medicine, McMaster University, Hamilton, Ontario. Dr Kronenberg is now affiliated with the Department of Medicine, University of Texas Medical Branch, Galveston. A complete list of the SOLVD investigators may be found in Bangdiwala SI, Weiner DH, Bourassa MG,

Collaborative Studies Coordinating Center, Department of Biostatistics, University of North Carolina at Chapel Hill (Ms Weiner and Dr Bangdiwala); the Division of Epidemiology and Clinical Applications, National Institutes of Health, Bethesda, Md (Dr Yusuf); the Departments of Medicine and Research, Brigham and Women's Hospital, Boston, Mass, the Veterans Affairs Medical Center, West Roxbury, Mass, and Harvard Medical School, Boston, Mass (Dr McIntyre); the Department of Medicine, Vanderbilt University School of Medicine, Nashville, Tenn (Dr Kronenberg); the Department of Medicine and Pharmacology, Robert Wood Johnson Medical School, New Brunswick, NJ (Dr Kostis); Departments of Medicine and Research, State University of New York at Buffalo (Dr Kohn); the Department of Medicine, Oregon Health Sciences

Woods);

APPROXIMATELY two million indi viduals in the United States are affected by congestive heart failure (CHF) and CHF is responsible for 900 000 hospitalizations and 37000 deaths annually.1 Reliable instruments to assess the se verity of symptoms, the degree of left ventricular dysfunction, prognosis, and response to therapy are essential for optimal clinical management of these pa tients and also in the evaluation of new therapies. The ideal instrument should be simple, noninvasive, inexpensive, and safe. Further, it should predict morbid ity and mortality in patients with CHF from varying causes and of varying se verity, and should provide unique data that complement already available noninvasive severity indexes and tests, such as the New York Heart Association (NYHA) functional heart failure class2 and ejection fraction (EF).S The 6-minute walk test has been used widely in the evaluation and serial as sessment of patients with chronic respi ratory disorders.45 Experience in CHF is more limited,6"9 but the test seems to be well accepted by patients, is easily implemented in an outpatient setting, and has good reproducibility.8 Correlations with NYHA class, specific activity scale, and standard bicycle ergometry are mod erate.9 It remains unclear whether this submaximal exercise test is prognostically useful. Patients enrolled in the Stud ies of Left Ventricular Dysfunction

University,

Portland (Ms Guillotte and Dr

Greenberg);

Friesinger GC II, Ghali JK, Yusuf S, for the SOLVD Investigators. Studies of Left Ventricular Dysfunction (SOLVD) Registry: rationale, design, methods and description of baseline characteristics. Am J Cardiol. 1992;70:347-353. Reprint requests to Department of Medicine, Division of Cardiovascular Disease, University of Alabama at Birmingham, THT 328-UAB Station, Birmingham, AL 35294 (Dr Bittner).

large, well-characterized subset of pa tients with left ventricular dysfunction from differing causes and of differing se verity who underwent the 6-minute walk test and were then followed up for a mean of 242 days. This report relates the dis tance walked in the 6-minute walk test to hospitalization and mortality rates dur
a

(SOLVD) Registry Substudy represent

ing follow-up.

Downloaded from www.jama.com at Capes Consortia on March 8, 2010

METHODS

ment in the main

ponents: (1) a randomized trial of enala-

Study Population The design of SOLVD has been de scribed in detail.10 In brief, SOLVD is a multicenter study with three major com

pulmonary edema, Kerley's lines, or pleural effusions judged secondary to CHF). Ejection fractions among walktest participants were measured by echocardiography in 90% of subjects, by radionuclide angiography in 7%, and by cineangiography in 3%. Over a 14-month period, 6273 patients were enrolled into

with no overt CHF and an EF of 0.35 or less (prevention trial); and (3) a registry of unselected patients with CHF or left ventricular dysfunction presenting to 20 hospitals in the United States, Canada, and Belgium (registry). The design of the SOLVD Registry and Registry Substudy has been de scribed in detail.11 Patients were eligible for enrollment if they had an E F of 0.45 or less and/or radiological evidence of CHF (one or more of the following: pul monary venous redistribution, basal or perihilar vascular blurring, alveolar or

pril malate vs placebo in symptomatic patients with CHF and an EF of 0.35 or less (treatment trial); (2) a randomized trial of enalapril vs placebo in patients

registry, the partici contacted by mail, by tele phone, or in person to ascertain vital sta tus and information on hospitalizations. The average time of follow-up from the date ofthe substudy visit (which was typi cally 3 months after entry into the main registry) was 242 82 days (mean SD); follow-up information was available in 895 (99.7%) of the 898 substudy patients. Records of hospitalizations were obtained from hospital medical records or refer ring physicians. Information about cir cumstances of death and death certifi cates were also obtained.
pants
were

endpoints of interest were (1) allmortality, (2) all hospitalizations, (3) hospitalizations for CHF, and (4) a com bined endpoint of death or hospitalization for CHF. Endpoint results were initially analyzed in terms of the proportion of patients in each quartile of walk-test per formance who experienced an event (quar tile 1, less than 308 m; quartile 2, 308 to 372 m; quartile 3,374 to 446 m; and quar
The
cause

the registry. Data were abstracted from the medical records of all registry par ticipants. The 898 patients participating in the SOLVD Registry Substudy rep resent a random sample of registry par ticipants stratified by cause of left ven tricular dysfunction with the following sampling proportions for the two strata: 36% for patients with ischemie heart dis ease and hypertensive heart disease and 100% for patients with left ventricular dysfunction due to other causes. The substudy patients underwent a 6-minute walk test and a detailed evaluation that in cluded a clinical examination, assessment ofNYHA functional class, a brief qualityof-life questionnaire, 12-lead electrocar diogram, chest roentgenogram, two-di mensional echocardiographic study, 70minute ambulatory electrocardiographic recording, and venipuncture for neurohormone measurements. Reference Group for the 6-Minute Walk Test.Forty age- and sex-matched individuals without history of left ven tricular dysfunction were recruited among patients' family members and friends and individuals known to the study staff. Dur ing the clinic visit, age, sex, height, weight, smoking history, and history of cardio vascular symptoms were ascertained and the 6-minute walk test and quality-of-life scales were administered. Follow-up of SOLVD Registry Substudy Patients.One year after enroll-

The 6-Mlnute Walk Test The 6-minute walk test was conducted as described by Guyatt et al.8 In a corri dor, a 30.5-m (100-ft) course was marked and a chair was placed at each end. Pa tients were instructed to walk from end to end at their own pace while attempt ing to cover as much ground as possible in the allotted period of 6 minutes. A study nurse timed the walk test, called out the time every 2 minutes, and encouraged the patient every 30 seconds in a stan dardized fashion (while facing the patient and using one of the two phrases "You're doing well" or " Keep up the good work"). Patients were allowed to stop and rest during the test, but were instructed to resume walking as soon as they felt they were able to do so. After 6 minutes had elapsed, patients were instructed to stop walking and the total distance walked was measured to the nearest meter (or

tile 4, greater than 446 m). At the request of the reviewers, endpoint results pre sented herein are for four performance levels using more intuitive cut points for distance walked during the 6-minute walk: level 1, less than 300 m; level 2, 300 to 374.9 m; level 3,375 to 449.9 m; and level 4,450 m or greater. The results ofthe two

analyses were comparable. Logistic regression assessed the change in odds of experiencing an endpoint as a
was entered as a continuous

function of distance walked when distance variable: odds ratios (ORs) express the increase in odds of an endpoint for a 120-m (approximately 1 SD) decrement in distance walked. Be cause selection into the substudy was stratified by cause, multivariate logistic regressions forced cause (ischemie and hypertensive heart disease vs other causes) into the models. Probability val ues are two-tailed and not adjusted for

multiple comparisons.

foot). Symptoms experienced by patients

Consent The SOLVD

during the walk (angina, dyspnea, fatigue, dizziness, and syncope) were recorded. Registry and Registry Substudy protocols were approved by the institutional review boards of all partici pating institutions and all patients en
rolled gave informed consent. Statistical Analysis Baseline comparisons of clinical char acteristics used Cochran-Mantel-Haenszel tests (for categorical variables) or lin ear models (for continuous variables) ad justing for clinic; because of small sample sizes, comparisons of test takers vs nonwalkers and test takers vs reference group used unadjusted tests. Relation ships between performance and clinical characteristics were tested by compar ing the mean distance walked (in meters) between clinical groups or, for continu ous variables, by Pearson correlations. When data were not normally distrib uted, nonparametric tests were used. In order not to obscure important baseline differences, probability values were not corrected for multiple comparisons.

RESULTS Patient Demographics and Clinical Characteristics Demographics and clinical character istics at the time of registry enrollment for patients in the registry (but not in the substudy) and patients in the registry substudy are given in Table 1. Substudy patients were younger, less likely to have ischemie or hypertensive heart disease (see "Methods" section), and less likely to take nitrates and calcium channel block ers. Despite a slightly lower EF, fewer substudy patients had pulmonary rales or edema on physical examination. Out of the 898 substudy patients, 833 (93%) participated in the 6-minute walk test (test takers) while 65 did not (nonwalkers). Reasons for nonparticipation in cluded inability to walk due to physical disabilities, patient refusal, and physician judgment. Proportionately fewer nonwhite patients participated in the walk test. Test takers were also older, ranked in a better NYHA class, less likely to take diuretics, and less likely to have rales on physical examination than were nonwalkers (Table 1). Subjects in the reference group had a significantly lower body mass index (meanSD, 24.73.1 kg/m2 vs 26.94.9 kg/m2; P=.004) and were less likely to have ever smoked (58% vs 76%; P=.02)

Downloaded from www.jama.com at Capes Consortia on March 8, 2010

Table 1.Baseline Clinical Characteristics

no

Registry (Not

in

(n=5375)*_(n=898)*
6312

Substudy) Substudy
5912 78/22
85/13

Test Takers Nonwalkers

(n=833)t
6012 78/22 86/12
3714

Age, y, meanSD
Sex, %, M/F Race, %, white/nonwhlte Ejection fraction, meanSD
Cardlothoraclc ratio on chest roentgenogram, %
S.52 >.52

(n=65)t 5614
68/32

73/27 86/12 31 9

66/32
3414

309

42

38 34
28

Missing
New York Heart Association class, %

15 35 47 14

40
31

23

Cause,H
Other

Ischemie heart disease 28

51
49

54

38

Other conditions present, % Chronic obstructive pulmonary disease Previous stroke

13

6*
12 22
12

14
15

Peripheral

vascular disease

Current smoker Medications, %

Distance walked decreased with age was shorter for women (330 vs 386 m; P<.0001) and black participants (319 vs 380 m; P<.0001). The correlation between E F and distance walked was weak. The relationship between YHA class and distance walked in performance lev els is shown in Fig 1. Clear trends for longer and shorter walking distances were evident for patients without symptoms ( YHA class I) and for those with severe symptoms (NYHA class III/IV), respec tively, although there was some overlap between the two groups. On the other hand, NYHA class II patients were rela tively evenly distributed across the four performance levels. Thus, while walk-test performance was consistent with func tional status at the extremes of the NYHA classification, it demonstrated a substan tial range in the broad class of moderate impairment (NYHA class II).

dyspnea.

symptoms, while six (14%) developed

(r= .34; P<.0001), and

Digitalis
Diuretics
58 59

58
15

-Blockers
Nitrates

16 46 37

Calcium blockers

36 31
32

32 36
12718

25 32
123 + 20

Anglotensin-convertlng enzyme inhibitors Physical examination Systolic blood pressure, meanSD

Diastolic blood pressure, meanSD Heart rate, beats/mln, meanSD

12822 7713
8217

12719

7712 8016
27.04.9

7711 7514 26.94.9

7510 7914 27.95.9

Body

mass

Index,

kg/m2,

mean SD

Elevated

jugular venous pressure,

21 34
15

20

Rales, % S3 gallop, % Mitral rgurgitation, % Edema, % Atrial fibrillation, % No signs of congestive

*Data obtained tData obtained

27
13

17
15 20
62

Morbidity and Mortality Fifty-two walk-test participants (6.2%) died during the follow-up period. Among the 52 deaths, 40 were cardiovascular, five were noncardiovascular, and seven could not be classified. During follow-up, 252 walk-test participants (30.3%) were hospitalized; 78 (9.4%) were hospitalized for CHF. The combined endpoint ofdeath or hospitalization for CHF occurred in 114 walk-test participants (13.7%). Among the nonwalkers, 6.2% died during the fol low-up period, 30.8% were hospitalized, and 10.8% were hospitalized for CHF.
The differences in event rates between test takers and nonwalkers were not sta

28
heart

18 22
15 51

13
72

tistically significant.

failure, %

49

registry enrollment. substudy enrollment. <.05 comparing substudy patients and registry patients not enrolled in the substudy. Substudy patients were younger and more likely to be male. Owing to the selection algorithm, they were less likely to have Ischemie heart disease and more likely to have a nonischemlc cause of left ventricular dysfunction; as expected, use of nitrates and calcium blockers was less common in these patients. Among substudy patients, ejection fraction was slightly lower and mitral rgurgitation more frequent, but rales and edema were less common and substudy patients were less likely to have suffered a prior stroke. P<.05 comparing test takers and nonwalkers. Proportionately fewer nonwhlte patients participated In the walk test. Walkers were older, were assigned to a lower New York Heart Association class, were less likely to take diuretics, had fewer past strokes, and were less likely to have rales on physical examination than nonwalkers. IIEIIIpses Indicate not available. IJCause of left ventricular dysfunction.

at the time of at the time of

to mortality (P<.02). Compared with those walking at least 450 m, patients walking less than 300 m had a 3.7-fold risk of dying (95% confidence interval [CI], 1.44 to 9.55), and those in level 2 had a 2.78-fold risk of dying (95% CI, 1.09 to 7.11); the risk of death was not significantly increased in patients in level 3 (OR, 1.42; 95% CI, 0.50 to 4.06). Mortality in the lowest level of

Distance walked was inversely related

compared with test takers. None were current smokers, and as expected, few gave a history of angina (5%), breathlessness (8%), or hypertension (10%). Walk-Test Performance in Substudy

Patients and Individuals in the Reference Group Among the 833 test takers, 49 patients were not able to complete the test, 43 com pleted the test with breaks in continuity, while the remaining patients completed the walk without stopping or resting. All

subjects in the reference group completed the test without breaks in continuity, and as expected, they walked further than substudy patients (mean SD, 555.2 114.5 m vs 374.3117.1 m; P<.0001). Most substudy patients (n=547, 66%) did not report any symptoms during the test; 41 patients (5%) developed angina, 206 (25%) complained of dyspnea, 194 (23%) had fatigue, and 32 (4%) reported dizziness. All symptoms resolved shortly after the completion of the walk. In the reference group, 86% of individuals had

distance walked was over three times that in the highest level (10.23% vs 2.99%) (Fig 2). The percentage of patients hospital ized for any reason and those hospitalized for CHF increased significantly as dis tance walked decreased (P<.001 for both endpoints) (Fig 3). There was no signifi cant relationship between distance walked and hospitalizations for noncardiovascu lar reasons. In contrast, compared with patients in the highest performance level (level 4), the odds of being hospitalized for CHF were strikingly increased in perfor mance levels 1 (OR, 14.02; 95% CI, 4.90 to 40.14) and 2 (OR, 6.21; 95% CI, 2.14 to

Downloaded from www.jama.com at Capes Consortia on March 8, 2010

Fig 1.Functional classification and distance walked. The patient population was divided into four performance levels based on distance walked. The percentage of patients in each performance level for each New York Heart Association functional heart failure class is shown. Note that Class II patients are evenly distributed among all four performance levels (see the "Results" section for additional information).

Fig 2.Mortality (%) as a function of performance level (based on distance walked). Mortality de creased as performance on the 6-minute walk test improved.

Fig 3.Hospitalizations as a function of perfor level (based on distance walked). Both hos pitalization rates decreased significantly as walking
mance

distance increased.

Fig 4.Combined endpoint of mortality or hospital ization for congestive heart failure as a function of performance level (based on distance walked) and ejection fraction. The relationship between distance walked and prognosis was significant within each ejection fraction subgroup (P<.01 and P<.001, re spectively, for an ejection fraction of 0.25 or less and an ejection fraction greater than 0.25).

Fig 5.Combined endpolnt of mortality or hospital ization for congestive heart failure as a function of performance level (based on distance walked) and New York Heart Association (NYHA) functional heart failure class. The relationship between distance walked and prognosis was significant in NYHA class I (P<.01) and NYHA class II (P-c.001), whereas in NYHA class lll/IV a nonsignificant trend was ob served (note that only a few patients among NYHA class lll/IV achieved performance levels 3 and 4).

95%

Figures 4 and 5 show the incidence of the combined endpoint of death or hos pitalization for CHF by performance level stratified by EF and NYHA class, re spectively. Within each stratum of E F
and among NYHA class I and NYHA class II patients, shorter walking distance was associated with a significantly higher risk of events during follow-up. This re lationship was less clear among NYHA class III/IV patients. Although distance walked tended to predict events better among patients with higher E F and bet ter NYHA class than in patients with more severe left ventricular dysfunction and more limited functional capacity, this trend did not reach statistical significance

18.08), with a trend in level 3 (OR, 1.90; CI, 0.56 to 6.42).

( 2 of distancexEF interaction, 1.17; df, 1; not significant; 2 ofdistance X NYHA interaction, 3.15; df, 2; not significant). In a stepwise logistic regression model

of hospitalization for CHF among NYHA class I and II pa tients (OR, 2.60; 95% CI, 1.78 to 3.80). An analysis among NYHA class III/IV pa tients was not feasible because ofthe small number of patients in this group. Little is known about prognostic de terminants among patients with CHF and preserved EF (EF greater than 0.45). Among our walk-test participants, 25.8% had radiological manifestations of CHF despite an EF greater than 0.45. Mortal ity during follow-up was slightly, but not significantly, lower among these patients compared with those with an EF of 0.45 or less (3.8% vs 6.9%; not significant), and they were hospitalized less frequently for CHF (5.2% vs 10.9%; P<.02). Owing to the low number of events in the group with preserved EF, distance walked as a predictor of prognosis in this subgroup could only be analyzed for the combined endpoint of death or hospitalization for CHF. Distance walked was an equally powerful predictor of this combined endpoint among patients with preserved EF as among those with impaired systolic left ventricular function with respective ORs (and 95% CIs) of 2.62 (1.57 to 4.37) and 1.99 (1.56 to 2.54) for a 120-m difference in distance walked.

pendent predictor

starting with age, sex, cause, NYHA class, and E F as covariates and eliminating non significant variables other than cause in subsequent iterations (Table 2), distance walked remained a significant indepen dent predictor of subsequent mortality (OR, 1.50 for each 120-m decrement in distance walked; 95% CI, 1.11 to 2.03) and the combined endpoint of death or hos pitalization for CHF (OR, 1.77; 95% CI, 1.38 to 2.26). After stratification by NYHA class because of a significant interaction
between distance walked and this covariate, distance walked was also an inde-

COMMENT This study demonstrates that the 6-minute walk test strongly and indepen dently predicts long-term mortality and hospitalization rates for CHF in patients with left ventricular dysfunction of vary ing severity and cause and among those with CHF and preserved EF. This single test provides prognostic information be yond that of traditional indicators such as EF and NYHA functional heart failure class. Moreover, the test can be admin istered safely in an outpatient setting

Downloaded from www.jama.com at Capes Consortia on March 8, 2010

Table 2.Predictors of 1-Year

Mortality, Hospitalization for Congestive Heart Failure,

Death

and Combined

Endpoint*
Death Odds
or Hospitalization for Congestive Heart Failure

Hospitalization for Congestive Heart Failuref

Odds Ratio 95% Confidence Interval
1.78-3.80

Variable

Odds Ratio

95% Confidence Interval 1.11-2.03 0.62-2.83

0.67-4.31

Ratio

95% Confidence Interval 1.48-2.49 1.38-2.26

0.70-2.11

Ejection fraction
Distance,
II
vs m

1.74 1.50$

2.55 2.60

1.92

1.77*
1.22 2.08

NYHA class

I
vs I
vs

M) Age, y Causelj (IHD or

lll/IV Sex (F

HHD

vs

other)

1.33 1.69 1.14 0.95 0.91

0.85
2.66 1.35 1.09

0.42-1.70 1.35-5.26 0.93-1.95

0.56-2.10

1.07-4.04 1.08-2.90
0.66-1.61

0.56-2.30 0.68-1.32 0.49-1.68

1.77*
1.11 1.03

*Odds ratios with 95% confidence intervals were calculated for a decrement of 1 SD of ejection fraction (-0.15) and walking distance (-120 m), for New York Heart Association (NYHA) class (lll/IV vs I and II vs I), sex (women vs men), an age difference of +12 years, and cause of left ventricular dysfunction (Ischemie heart disease [IHD] or hypertensive heart disease [HHD] vs other). tBecause of a significant interaction between walking distance and NYHA class, analysis was restricted to NYHA class I and li; there were too few NYHA class lll/IV patients for complete stratification. 4These odds ratios are from the stepwise model, which Included distance walked, cause of left ventricular dysfunction, and any significant covariates. These odds ratios are for the nonsignificant variables dropped In the stepwise procedure; values for these variables, obtained from the full model, are reported for comparative
purposes.

(Cause

of left ventricular

dysfunction.

without the need for specialized equip ment and is well accepted by patients of varying ages and debilitation. Determinants of Walk-Test Performance Distance walked in this study discrimi nated well between a healthy reference group and age- and sex-matched individu als with left ventricular dysfunction, which is consistent with previous work by Lipkin et al.7 Among the CHF patients, those with worse NYHA functional class gen erally walked shorter distances than in dividuals with better NYHA class, but the association between this measure of disease severity, albeit highly statistically significant because of large sample sizes, was only small in magnitude. A similar moderate correlation between perfor mance on the 6-minute walk test and NYHA class has also been observed by others,7,9 suggesting that these two mea sures assess different aspects offunctional status. Similarly weak correlations are seen between NYHA class and maximal exercise capacity as determined by tra ditional symptom-limited or cardiopulmonary exercise testing.12"14 E chocar diographic left ventricular di mensions and radionuclide left ventricu lar EF measured at rest correlate poorly with maximal exercise capacity.1417 In formation about determinants of submaximal exercise is limited to hemodynamic indexes,1819 and determinants of the 6-minute walk test have not been previously evaluated. In our study, dis tance walked on the 6-minute walk test was only weakly related to resting left ventricular EF. Prediction of Morbidity and Mortality This is the first study to demonstrate a strong relationship between perfor
mance on a

tients with left ventricular

dysfunction. Mortality during follow-up was rela tively low in our population compared with other published series of patients with CHF.*"** Three factors are likely to be responsible. First, we had very few pa

test and

simple, self-paced walking morbidity and mortality in pa-

tients who were classified as NYHA class IV, and 36% of walk-test participants were in NYHA class I. Second, 25.8% of substudy participants qualified for the study based on radiological evidence of CHF and had an EF greater than 0.45. Heart failure due to diastolic dysfunction is known to carry a better prognosis.23 Third, 36% of substudy patients were tak ing angiotensin-converting enzyme inhibi tors, which are known to improve sur vival in patients with CHF.2425 In a model including cause, age, sex, EF, NYHA class, and walking distance, E F and walking distance were the only significant independent predictors of mor tality. Despite some evidence to the con trary,1526 most studies have shown that maximal oxygen consumption determined by bicycle or treadmill exercise is an in dependent predictor of mortality in pa tients with CHF.1617202127 However, these studies require specialized equipment and trained staff. Therefore, there is a need for a simpler test that can be performed without expensive equipment in most set tings, including a physician's office. Our study documents for the first time that low-level submaximal exercise testing can predict mortality in patients with left ven tricular dysfunction. While patients with severe CHF (classes III and IV) are known to have a high mortality, it has not been previously possible to reliably dis criminate high-risk subgroups among the heterogeneous pool of patients with mild CHF. Our study demonstrates that the 6-minute walk test is a particularly useful discriminator of outcome among patients with mild to moderate CHF or asymp tomatic left ventricular dysfunction. Table

3 shows how walk-test results can "riskstratify" patients presenting with mild to moderate CHF. We are not aware of any previous study that has related any measure of func tional capacity to hospitalization rates. In our study, performance on the 6-minute walk test was an independent predictor of risk of hospitalization over the subse quent year in patients with left ventricu lar dysfunction due to ischemie or hyper tensive heart disease. The relationship between distance walked and hospital ization rate for CHF was even more strik ing. Each decrement in distance walked of 120 m (approximately one-half city block) resulted in a 160% increase in risk of hospitalization for CHF. Only EF and sex had a similarly strong relationship with CHF hospitalization rates. Thus, dis tance walked during the 6-minute walk test is a powerful long-term predictor of hospitalization for CHF in patients with left ventricular dysfunction and provides information above and beyond traditional indicators of prognosis such as EF and NYHA functional class.
Ease of Implementation We did not encounter any significant complications in over 800 walk tests in this study. The 6-minute walk was well accepted by individuals of both sexes and all age groups. Even quite debilitated pa tients did not feel threatened by the test, and many managed to complete the walk despite breaks in continuity due to symp toms. This self-paced exercise test more closely mimics the patients' day-to-day activities than maximal exercise testing. Strict standardization is essential for re

Safety and

producible results; study personnel were, however, trained in a short period. There fore, the test could be easily incorporated into routine outpatient visits in clinical practice. Since our study was performed in 20 hospitals in three countries, its re-

Downloaded from www.jama.com at Capes Consortia on March 8, 2010

Table 3.Refinement of as Predicted by Model*

Prognosis by Results From the 6-Minute Walk: Examples of Probabilities of Events

Typical Mortality
.066 Man

rax.

Prognostic Ejection fraction, 0.35;

Variables
New

for Congestive Heart Failure

Hospitalization

Typical
Mortality

Woman

for Congestive Heart Failure

Hospitalization

York Heart Association class II; patient walked 300 m York Heart Association class II; patient walked 375 m New York Heart Association class II; patient walked 450 m

.064

Ejection fraction, 0.35; New Ejection fraction, 0.35;

.054

.044

.061

.089

.045

.030

.050

.062

*Probablities of dying or being hospitalized for heart failure are calculated for a "typical" patient aged 50 years with an ejection fraction of 0.35 and New York Heart Association class II symptoms. Because of study design, probabilities must be calculated separately for Ischemie or hypertensive heart disease and other causes. Probabilities reported here are for the ischemie heart disease category; probabilities for hypertensive heart disease were comparable. Significant heterogeneity in prognosis is evident among both women and men with different walk-test

performances.

suits are likely to be more broadly appli cable than the results of studies from one or two specialized centers.

It is thus unlikely that additional walks would have materially altered our results.

Limitations The SOLVD Registry Substudy popu lation is a heterogeneous sample of pa tients with both systolic and diastolic left ventricular dysfunction from varying causes. Compared with the general CHF population, unusual causes were overrepresented by our study design; but, due to the large sample size, this relative overrepresentation should not limit the ap plicability of our results to clinical prac tice. Furthermore, cause of CHF was not a significant determinant ofwalk-test per formance. The ability to show a clear re lationship with morbidity and mortality despite the inclusion of patients with di verse causes of left ventricular dysfunc tion adds to the clinical value of the test. Sixty-five patients did not participate in the walk test. Even though there were some statistically significant differences in baseline characteristics between test takers and nonwalkers, the similarity in hospitalization and mortality rates sug gests that the exclusion of these patients is unlikely to have significantly biased the results of our study. Walk tests were performed by a vari ety of staff members in our clinical cen ters and reproducibility was not formally assessed. Despite standardization of the procedure for administration of the 6-minute walk test to maximize consis tency among centers, considerable het erogeneity in patient performance among centers was apparent, which was at least in part related to differences in the local patient populations. Patients were only tested once according to our protocol. It is possible that absolute distances walked might have increased with subsequent walks due to the "training effect" previ ously documented.6 Even though distance may increase during multiple testing, first and second walks are highly correlated.28

Conclusion The 6-minute walk test is a safe and simple clinical tool that independently relates to morbidity and mortality in patients with left ventricular dysfunc tion and in patients with CHF, despite preserved EF. It is particularly useful in identifying patients with higher event rates among the large group of patients with mild to moderate symptoms of CHF (NYHA class II). As is true for other prognostic markers, results of the 6-minute walk test should not be ap plied clinically without consideration of the entire patient profile. Future stud ies should explore the usefulness of the 6-minute walk in serially monitoring therapy and progression of left vent ricular dysfunction.
This study was supported by contract NOI HC55001 from the National Institutes of Health, Bethesda, Md.
References 1. Kannel WB, Belanger AJ. Epidemiology of heart failure. Am Heart J. 1991;121:951-957. 2. Braunwald E. Examination of the patient. In: Braunwald E, ed. Heart Disease: A Textbook of Cardiovascular Medicine. 4th ed. Philadelphia, Pa: WB Saunders Co; 1992:11-12. 3. Franciosa JA, Park M, Levine TB. Lack of correlation between exercise capacity and indexes of resting left ventricular performance in heart failure. Am J Cardiol. 1981;47:33-39. 4. Butland RJA, Pang J, Gross ER, Woodcock AA, Geddes DM. Two, six, and 12 minute walking tests in respiratory failure. BMJ. 1982;284:1607-1608. 5. Knox AJ, Morrison JFJ, Muers MF. Reproducibility of walking test results in chronic obstructive airways disease. Thorax. 1988;43:388-392. 6. Guyatt GH, Sullivan MJ, Thompson PJ, et al. The 6 minute walk: a new measure of exercise capacity in patients with chronic heart failure. Can Med Assoc J. 1985;132:919-923. 7. Lipkin DP, Scriven AJ, Crake T, Poole-Wilson PA. Six minute walking test for assessing exercise capacity in chronic heart failure. BMJ. 1988;292:653-655. 8. Guyatt GH, Pugsley SO, Sullivan MJ, et al. Effect of encouragement on walking test performance. Tho-

9. Guyatt GH, Thompson PJ, Berman LB, et al. How should we measure function in patients with chronic heart and lung disease? J Chronic Dis. 1985;38:517\x=req-\ 524. 10. The SOLVD Investigators. Studies of left ventricular dysfunction (SOLVD): rationale, design and methods: two trials that evaluate the effect of enalapril in patients with reduced ejection fraction. Am J Cardiol. 1990;66:315-322. 11. Bangdiwala SI, Weiner DH, Bourassa MG, Friesinger GC, Ghali JK, Yusuf S, for the SOLVD Investigators. Studies of Left Ventricular Dysfunction (SOLVD) Registry: rationale, design, methods and description of baseline characteristics. AmJCardiol. 1992;70:347-353. 12. Jafri SM, LakierJB, Rosman HS. Symptoms and tests of ventricular performance in the evaluation of chronic heart failure. Am Heart J. 1986;112:194-196. 13. Franciosa JA, Ziesche S, Wilen M. Functional capacity of patients with chronic left ventricular failure. Am J Med. 1979;67:460-466. 14. Dunselman PHJM, Kuntze CEE, van Bruggen A, et al. Value of New York Heart Association classification, radionuclide ventriculography, and cardiopulmonary exercise tests for selection of patients for congestive heart failure studies. Am Heart J. 1988; 116:1475-1482. 15. Baker BJ, Leddy C, Galie N, Casebolt P, Franciosa JA. Predictive value of M-mode echocardiography in patients with congestive heart failure. Am Heart J. 1986;111:697-702. 16. Szlachcic J, Massie BM, Kramer BL, Topic N, Tubau J. Correlates and prognostic implications of exercise capacity in chronic congestive heart failure. Am J Cardiol. 1985;55:1037-1042. 17. Cohn JN, Rector TS. Prognosis of congestive heart failure and predictors of mortality. Am J Cardiol. 1988;62:25A-30A. 18. Gibbs JSR, Keegan J, Wright C, Fox KM, Poole\x=req-\ Wilson PA. Pulmonary artery pressure changes during exercise and daily activities in chronic heart failure. J Am Coll Cardiol. 1990;15:52-61. 19. Meiler SEL, Ashton JJ, Moeschberger ML, Unverferth DV, Leier CV. An analysis of the determinants of exercise performance in congestive heart failure. Am Heart J. 1987;113:1207-1217. 20. Willens HJ, Blevins RD, Wrisley D, Antonishen D, Reinstein D, Rubenfire M. The prognostic value of functional capacity in patients with mild to moderate heart failure. Am Heart J. 1987;114:377-382. 21. Cohn JN, Archibald DG, Francis GS, et al. Veterans Administration Cooperative Study on Vasodilator Therapy of Heart Failure: influence of prerandomization variables on the reduction of mortality by treatment with hydralazine and isosorbide dinitrate. Circulation. 1987;75(suppl IV):49-63. 22. Gradman A, Deedwania P, Cody R, et al. Predictors of total mortality and sudden death in mild to moderate heart failure. J Am Coll Cardiol. 1989; 14:564-570. 23. Cohn JN, Johnson G, Veterans Administration Cooperative Study Group. Heart failure with normal ejection fraction: the V-HeFT Study. Circulation.
24. The SOLVD Investigators. Effect of enalapril on survival in patients with reduced left ventricular ejection fractions and congestive heart failure. N Engl J Med. 1991;325:293-302. 25. Cohn JN, Johnson G, Ziesche S, et al. A comparison of enalapril with hydralazine-isosorbide dinitrate in the treatment of chronic congestive heart failure. N Engl J Med. 1991;325:303-310. 26. Franciosa JA. Why patients with heart failure die: hemodynamic and functional determinants of survival. Circulation. 1987;75(suppl IV):20-27. 27. Griffin BP, Shah PK, Ferguson J, Rubin SA. Incremental prognostic value of exercise hemodynamic variables in chronic congestive heart failure secondary to coronary artery disease or to dilated cardiomyopathy. Am J Cardiol. 1991;67:848-853. 28. Rozkovec A, Papouchado M, James MA, Kendrick AH, Clarke LM, Rees LR. The relationship of symptoms to performance in paced patients with breathlessness. Eur Heart J. 1989;10:63-69.

1984;39:818-822.

1990;81(suppl III):48-53.

Downloaded from www.jama.com at Capes Consortia on March 8, 2010