Professional Documents
Culture Documents
ABSTRACT Background and Purpose. To assess the reproducibility and validity of the
six-minute walk test (6MWT) in men and women with obesity in order to facilitate evalua-
tion of treatment outcome. Method. A test—retest design was used to test reproducibility
and a comparative design to test known group validity. Forty-three obese outpatients (16
male), mean age 47 (21–62) years, mean body mass index (BMI) 40 (3–62)kg-m–2 per-
formed the 6MWT twice within one week. Intraclass correlation (ICC1.1) and measurement
error (Sw) were calculated from the mean square values derived from a one-way repeated-
measures ANOVA (fixed effect model). The reproducibility was also analysed by means of
coefficient of variation (CV) and the Bland Altman method including 95% limits of agree-
ment. The variance of the distance walked was analysed by means of regressions. The
known group validity of the 6MWT (distance walked and the work of walking) in obese
participants was shown by comparisons with 41 lean participants (18 male), mean age 47
(24–65) years, mean BMI 22.7kg-m–2 (19–25). Results. The obese group walked 534 m
(confidence interval [CI] 508–560 the first and 552 m (CI 523–580) the second walk (p <
0.001). Sw was 25 m, CV 4.7%, ICC1.1 was 0.96. The limits of agreement were –46 m+80 m.
The validity tests showed that they walked 162 m shorter (p < 0.001) and performed much
heavier work (p < 0.001) than the lean group. In the obese group, BMI alone explained
38% of the variance of the distance walked. Conclusions. The 6MWT showed good repro-
ducibility and known group validity and can be recommended for evaluating walking ability
in subjects with obesity. For individual evaluation, however, an improved walking distance
of at least 80 m was required to make the difference clinically significant. Despite shorter
walking distance the obese participants performed heavier work than the lean. Copyright
© 2008 John Wiley & Sons, Ltd.
obese group were compared with results Anthropometric data on obese and lean
from one test in a lean group. participants are presented in Table 1. Age
and height did not differ significantly, but
Subjects weight and BMI did (p < 0.001).
Informed consent was obtained from all
For the reproducibility testing, male and participants and the study was approved by
female outpatients registered at The Centre the local ethics committee.
for Obesity Treatment, Norrtulls Hospital,
Karolinska University Hospital, Stockholm Methods and procedure
and Sweden, were during six months con-
secutively recruited and invited to partici- The obese participants performed the 6MWT
pate in the study. Patients with known severe twice with a median interval of 5 (range 2–
psychiatric diagnoses, difficulty in under- 14) days. They were instructed to wear com-
standing Swedish, walking aids or total hip fortable shoes and clothes. Each participant
replacements were excluded. Forty-three was tested at the same time in the morning
patients chose to participate. Seven were and by the same physiotherapist at both
medicating with beta-blockers and all medi- occasions.
cation was stable during the study period. The 6MWT was performed according to
Their mean age was 46.8 (21–62) years and ATS’ (2002) guidelines in an indoor corri-
mean body mass index (BMI) 39.6 (30.4– dor. Two cones 40 m apart indicated the
59.1)kg-m–2. Those 43 patients who declined length of the walkway. The instructions
participation due to tight working schedule, were: ‘You shall now walk as far as possible
vacation or unknown reason did not sig- for six minutes, but don’t run or jog’. Stan-
nificantly differ from those who chose dardized encouragement was given every
participation. minute by telling how many minutes were
For the validation process of the 6MWT left. Heart rate (HR) before (at rest) and
in the obese group, reference values were directly after the 6MWT was palpated at the
collected from lean volunteers, recruited by wrist. A stopwatch (Sport Timer, K-Tech R)
flyers posted at the hospital. Forty-one, mean was used for timekeeping and HR measures.
BMI 22.7 (19–25)kg-m–2, aged 46.5 (24–65) Directly after the tests the participants were
years, volunteered to participate. asked to rate perceived exertion and/or pain
Age (years) 43.9 (39.2–48.5) 45.2 (40.1–50.3) 51.8 (47.1–56.4) 47.9 (42.0–54.1)
Body mass (kg) 107.8 (101.7–113.8) 61.9 (58.6–65.2) 126.0 (113.8–138.2) 76.1 (73.1–79.0)
Height (m) 1.66 (1.63–1.69) 1.67 (1.65–1.70) 1.78 (1.75–1.82) 1.80 (1.77–1.82)
BMI 39.4 (37.1–48.5) 22.1 (21.3–22.8) 39.9 (35.8–44.0) 23.6 (22.9–24.2)
Mean (95% confidence interval). Obese (n = 43), and lean participants (n = 41).
BMI = body mass index.
ICC1.1
on the Borg CR10-scale (Borg, 1982). For
0.96
the validation, one 6MWT in the lean group
Sw = measurement error; CV = coefficient of variation; SDdiff = standard deviation of the differences; ICC1.1 = intraclass correlation coefficient.
was sufficient to make comparisons with the
first walk in the obese group.
SDdiff
Weight was measured to the nearest
31.6
0.1 kg on an electronic scale (Vetek ‘Unis-
cale’, Vetek AB, SE-760 40 Sweden) and
CV
4.7
height was measured to the nearest 0.5 cm
with a wall-mounted stadiometer (Seca-220,
25.3
701/702). Weight and height were then used
Sw
to calculate BMI.
agreement
Statistical analysis
Mean (95% confidence interval) or median (range). P calculated with ANOVA or Sign test (ordinal data).
56
47
%
Since beta-blockers could affect the HR
response and thus the sub maximal work and
<0.001
=0.92
=0.65
distance walked, an analysis of differences
=1.0
between obese participants with and without
p
this medication was performed by means of
Systematic
unpaired t-tests. Descriptive statistics are
difference
TABLE 2: Reproducibility of the distance walked in the obese participants (n = 43)
presented as mean (standard deviation [SD])
17.6
or 95% confidence interval (CI) or median
and range. The methods for assessing the
reproducibility of the 6MWD are presented 125 (118.6–131.4)
551.5 (523–580)
100
Limit of
80 agreement
(mean+2SD)
60
40
diff (m)
20
Bias
0 (mean)
-20
-40 Limit of
agreement
-60 (mean-2SD)
0 200 400 600 800
mean distance walked (m)
FIGURE 1: Agreement between two tests of distance walked in six-minute walk test.
means of a Bland-Altman plot (Bland and All but one obese participant could com-
Altman, 1986). plete the 6MWT without stopping for breaks
Known group validity (Finch et al., 2002) due to pain and/or dyspnoea. Twenty-eight
of the 6MWT and 6MWORK (kg·km) was others perceived musculoskeletal pain and
analysed by comparing results from obese all 43 perceived exertion. The distance
and lean participants with unpaired t-tests or walked was 17.6 m longer at the second test
the Mann-Whitney U-test. For identification (p < 0.001), but neither final HR, perceived
of variables (Table 3) with large impact on pain nor exertion differed between the tests.
the variance of 6MWD in the patients, first Analyses of reproducibility of the 6MWD
univariate and then backwards multiple showed that r was 0.94, Sw was 25.3 m, CV
regressions were used. was 4.7%, SDdiff was 31.6 m and ICC1.1
All comparisons are based on results was 0.96 (Table 2). The agreement between
from test 1 if nothing else is stated. All anal- the two tests is illustrated with a Bland
yses were performed in Microsoft® Office Altman plot in Figure 1. Considering bias,
Excel 2003 and StatSoft, STATISTICA 6.0. the limits of agreement were –45.6 m and
79.5 m.
RESULTS Twelve lean participants perceived pain
and all 41 perceived exertion. They walked
Results from the 6MWT are presented in 161.5 m further (p < 0.001), perceived less
Table 2, Figure 1 (reproducibility) and Table pain and exertion (p < 0.001), but their final
3 (validity). The seven obese and two lean HR did not differ (p = 0.24) from that of the
participants who medicated with beta-block- obese participants. 6MWORK was heavier
ers are included since beta-blockers did not in obese than in lean participants (p < 0.001;
influence the results. Since the medication Table 4).
may affect the HR response, the HR values In the obese group, women walked 531
of these nine participants were excluded (76) m and men 539 (89) m (p = 0.77).
from the analyses. 6MWD correlated negatively with BMI
TABLE 3: Backwards multiple regression between the second distance walked and BMI, fi nal heart rate, heart rate at rest, height and age in the obese
=0.002
(r = –0.63), pain (rs = –0.28) and age (r =
<0.001
–0.16) and positively with height (r = 0.44),
p
final HR (r = 0.35) and exertion (rs = 0.17).
Multiple regressions in steps showed that the
0.54
1.47
variables BMI, HR, height and age could
R2 = adjusted amount of explanation; B = regression coefficient; Std err = standard error of B; HR = heart rate; BMI = body mass index.
explain 71% of the variances in the second
R 2 = 0.52
–8.20
1.81
656.41
Step 4
(Table 3).
B
DISCUSSION
=0.003
<0.001
<0.001
0.57
1.35
0.77
=0.01
1.27
0.52
0.72
95.16
–6.22
202.88
2.77
250.76
–2.17
=0.08
=0.01
0.77
0.71
95.75
–1.90
301.03
–1.42
2.47
–6.75
Step 1
HR at rest
n = 36.
Height
Age
TABLE 4: Results from the fi rst six-minute walk test in obese (n = 43) and lean participants (n = 41)
Mean (95% confidence interval) or median (range); p calculated with unpaired t-test and Mann-Whitney U
test.
HR = heart rate.
about reactions to the fi rst test. In a few was required to be of clinical significance.
cases it became apparent that the partici- Such an improvement may seem large, but
pants did not fully understand the instruc- our results are comparable with others.
tions given the first walk. According to ATS’ Reported improvements of individual clini-
guidelines they were instructed to walk as cal importance amount to 65 m–133 m or
far as possible, which could have resulted in 10%–23% (Redelmeier et al., 1997; Enright,
slightly restricted speed. Some studies have 2003; Li et al., 2005; Andersson et al., 2006;
avoided this problem by instructions to walk Cunha et al., 2006).
as fast as possible (Gibbons et al., 2001; Concerning the known group validity we
Hulens et al., 2003; Sarsan et al., 2006), found a difference between the groups in
which might pose a risk of over-exertion at 6MWD, 6MWORK, pain and exertion. The
an early stage of the 6MWT. It is well known influence of BMI on the variance in 6MWD
that familiarisation with the 6MWT, not just was reinforced by the multiple regression
one but several times, may increase the analysis within the obese group, where the
6MWD and the total learning effect by as amount of explanation was 38%. Continued
much as 43 m (Troosters et al., 1999; Gibbons analyses revealed that obese participants
et al., 2001; Andersson et al., 2006). The with BMI >41kg-m–2, walked 84 m shorter
recommendations from the ATS suggest that distance (p < 0.001) than those with less
the first 6MWD results at baseline may be obesity. As 6MWD did not differ between
used for future outline comparisons. The genders, our results confirm the validity of
rationale given is that the 6MWD is only the 6MWT for both men and women with
slightly higher than the second test. This obesity.
seems sensible for clinical settings, since it The obese group in the present study
would be far too time-consuming and expen- walked 534 (285–716) m and covered only
sive to test patients on more than one occa- 77% of the distance of the lean group. Our
sion to get a reliable baseline value. results are comparable with the study by
A statistically significant mean increase Hulens et al. (2003), where women with
in 6MWD in a group is often much less than BMI >35kg-m–2 walked on average 539 (68)
a clinically significant increase in an indi- m and covered 75% of the distance of lean
vidual (ATC, 2002). For individuals in our women. In the studies by Maniscalco et al.
sample of obese participants an improve- (2006) and Sarsan et al. (2006) women
ment of the 6MWD of at least 80 m or 15% with obesity walked from 380 to 580 m,
depending on obesity level. As pain condi- will have no significant impact on the
tions might well interfere with walking 6MWD.
capacity (Mattsson et al., 1997; Evers We used several statistical methods, such
Larsson and Mattsson, 2001; Hulens et al., as the ICC and Bland and Altman tests,
2003) we had expected pain to limit the to analyse the reproducibility. ICC-values
6MWD in our obese participants to a higher depend on measurement precision and vari-
extent. More than 65% of them rated pain, ance between subjects, whereas Sw depends
ranging from just noticeable (0.5) to very on variance within subjects. We also pre-
strong (7) on the CR10-scale. Most pain sented the CV as it is expressed in percent-
emanated from knees or tibial periosteum. age, which enables comparisons with other
In the obese group BMI, HR, height and studies (Gulmans et al., 1996; ATC, 2002;
age explained 71% of the variance of the Wu et al., 2003). Altogether, these statistical
6MWD. BMI was the most important con- methods complemented each other and yet
tributor, whereas gender and pain unexpect- showed similar results. This will strengthen
edly did not influence the 6MWD. However our conclusion.
other factors might be of importance. It Also the bias between the two tests of
seems probable that low cardio respiratory 6MWD will have to be considered. Others
fitness would result in shorter 6MWD in the have avoided bias as they used a second test
present study. Previous studies on obesity as baseline value for outcome evaluations
have shown that low aerobic capacity in and they recommend test familiarisation
women resulted in high oxygen cost during (Gibbons et al., 2001; Wu et al., 2003;
walking (Mattsson et al., 1997; Browning Andersson et al., 2006). Since prevailing cir-
and Kram, 2005) and could explain 11% of cumstances in most clinical settings will not
the variance of the 6MWD (Hulens et al., make trial walks possible, we relied on the
2003). Calculations of 6MWORK showed statement by ATS and deliberately chose
that our obese participants performed much two tests to obtain clinically relevant mea-
heavier work during walking than our lean sures and, to put focus on the size of the Sw
participants. To use the 6MWORK as a between the first two tests. We believe that
parameter for individual fitness in patients our results will thus be applicable in patients
with obesity could be a simple method to similar to our participants for evaluation of
estimate the work of walking and functional intervention outcomes in individuals or in
capacity if no better methods are available groups.
(Chuang et al., 2001; Carter et al., 2003).
There may be some limitations in our IMPLICATIONS
study. We believe the results can be general-
ized since all test methods are standardized. Just one 6MWT should be sufficient to eval-
However, Swedes living in big cities are uate walking ability and functional capacity
accustomed to walking, thus the results may in the clinic in male and female patients with
not be generalized to obese subjects in all obesity. We can expect obese patients to
countries and cultures. Further, the length of walk significantly shorter distances and yet
our walkway was 40 m. ATS recommends perform heavier work than lean subjects.
30 m, but at the same time states that as long However, clinicians must be aware of the
as the walkway is straight and limited to size of the Sw. For individual evaluation over
between 15 m and 50 m, the walkway length time or after intervention the improvement
of the distance walked has to be approxi- adolescents with cystic fibrosis. Pediatric Pulm-
mately 15% of the baseline value or 80 m to onology 2006; 41: 618–622.
Domholdt E. Physical Therapy Research (second
be significant. edition). Philadelphia, PA: Saunders, 2000.
Donnelly JE, Jacobsen DJ, Jacicic JM, Whatley J,
ACKNOWLEDGEMENTS Gunderson S, Gillespie WJ, Blackburn GL, Tran
ZV. Estimation of peak oxygen consumption from
We are much obliged to all patients and lean reference
a sub-maximal half mile walk in obese females.
subjects for their participation, as to Elisabeth Berg,
International Journal of Obesity 1992; 16:
statistician for valuable advice, and Anne Christensson
585–589.
and Maren Schuldt, physiotherapists for expert help
Enright PL. The six-minute walk test. Respiratory
with the testing and clinical advice.
Care 2003; 48: 783–785.
The study was supported by the Committee for
Evers Larsson U, Mattsson E. Functional limitations
the Health and Caring Sciences at the Karolinska
linked to high body mass index, age and current
Institute.
pain in obese women. International Journal of
Obesity and Related Metabolic Disorders 2001;
REFERENCES 25: 893–899.
Finch E, Brooks D, Stratford PW, Mayo NE. Physical
American Thoracic Society (ATC). ATS statement: rehabilitation outcome measures. A guide to
guidelines for the six-minute walk test. American enhanced clinical decision making. Canadian
Journal of Respiratory and Critical Care Medi- Physiotherapy Association. Hamilton: Lippincott,
cine 2002; 166: 111–117. Williams & Wilkins, 2002.
Andersson C, Asztalos L, Mattsson E. Six-minute Gibbons WJ, Fruchter N, Sloan S, Robert D. Refer-
walk test in adults with cerebral palsy. A study of ence values for a multiple repetition 6-minute
reliability. Clinical Rehabilitation 2006; 20: walk test in healthy adults older than 20 years.
488–495. Journal of Clinical Rheumatology 2001; 21:
Bland JM, Altman DG. Statistical methods for assess- 87–93.
ing agreement between two methods of clinical Gulmans VA, van Veldhoven NH, de Meer K, Helders
measurement. Lancet 1986; 8: 307–310. PJ. The six-minute walking test in children with
Bland JM, Altman DG. Measurement error. British cystic fibrosis: reliability and validity. Pediatric
Medical Journal 1996; 313: 744. Pulmonology 1996; 22: 85–89.
Borg GAV. Psychosocial bases of perceived exertion. Hulens M, Vansant G, Claessens AL, Lysens R, Muls
Medicine and Science in Sports and Exercise E. Predictors of 6-minute walk test results in lean,
1982; 14: 377–381. obese and morbidly obese women. Scandinavian
Browning RC, Kram R. Energetic cost and preferred Journal of Medicine and Science in Sports 2003;
speed of walking in obese vs. normal weight 13: 98–105.
women. Obesity Research 2005; 13: 891–899. Li AM, Yin J, Yu CCW, Tsang T, So HK, Wong E,
Carter R, Holiday DB, Nwasuruba C, Stocks J, Chan D, Hon EKL, Sung R. The six-minute walk
Grothues C, Tiep B. 6-minute walk work for test in healthy children; reliability and validity.
assessment of functional capacity in patients with The European Respiratory Journal 2005; 25:
COPD. Chest 2003; 123: 1408–1415. 1057–1060.
Chen KY, Acra SA, Donahue CL, Sun M, Buchowski Maniscalco M, Zedda A, Giarddiello C, Faraone S,
MS. Efficiency of walking and stepping: relation- Cerbone MR, Cristiano S, Sofia M. Effect of bar-
ship to body fatness. Obesity Research 2004; 12: iatric surgery on the six-minute walk test in severe
982–989. uncomplicated obesity. Obesity Surgery 2006; 16:
Chuang M-L, Lin I-F, Wasserman K. The body 836–841.
weight-walking distance product as related to Mattsson E, Evers Larsson U, Rössner S. Is walking
lung function, anaerobic threshold and peak VO2 for exercise too exhausting for obese women?
in COPD patients. Respiratory Medicine 2001; International Journal of Obesity and Related
95: 618–626. Metabolic Disorders 1997; 21: 380–386.
Cunha TM, Rozov T, Caitano de Oliveira R, Jardim Rankin G, Stokes M. Reliability of assessment tools
JR. Six-minute walk test in children and in rehabilitation: an illustration of appropriate
statistical analyses. Clinical Rehabilitation 1998; Weir JP. Quantifying test-retest reliability using
12: 187–199. intraclass correlation coefficient and the sem.
Redelmeier DA, Bayoumi AM, Goldstein RS, Guyatt Journal of Strength and Conditioning Research
GH. Interpreting small differences in functional 2005; 19: 231–240.
status: the six minute walk test in chronic lung World Health Organization (WHO). International
disease patients. American Journal of Respiratory classification of functioning, disability and health
and Critical Care Medicine 1997; 155: 1278– (ICF). Geneva: WHO, 2001. (Available at: www3.
1282. who.int/icf/icftemplate.cfm) (Accessed 20 August
Sarsan A, Ardic F, Özgen M, Topuz O. The 2006).
effects of aerobic and resistance exercices in Wu G, Sanderson B, Bittner V. The 6-minute walk
obese women. Clinical Rehabilitation 2006; 20: test: how important is the learning effect?
773–782. American Heart Journal 2003; 146: 129–133.
Stucki A, Daansen P, Fuessi M, Cieza A, Huber E,
Atkinson R, Kostanjsek N, Stucki G, Ruof J. ICF Address correspondence to: Ulla Evers Larsson,
core sets for obesity. Journal of Rehabilitation Karolinska Institutet, Dept of Neurobiology, Care
Medicine 2004 Suppl; 44: S107–S113. Sciences and Society, Division of Physiotherapy 23
Troosters T, Gosselink R, Decramer M. Six minute 100, S-141 83 Huddinge, Sweden. E-mail: Ulla.Evers.
distance walked in healthy elderly subjects. Larsson@ki.se
The European Respiratory Journal 1999; 14: 270–
274. (Submitted June 2007; accepted January 2008)