Upsala J Med Sci 82: 27-30, 1977

On the Reproducibility of Exercise Tests in
Patients with Atrial Fibrillation

HANS &ERG, GUNNAR STROM and WAR WERNER

From the Departments of Internal Medicine and Clinical Physiology,
University Hospital, Uppsala, Sweden

ABSTRACT
Fifteen patients with atrial fibrillation, mostly due to
advanced valvular heart disease, were examined by a
graded work test and an orthostatic test on two occasions,
the reason being, to study the reproducibility of the test in
such patients. In 6 of the 7 patients with the highest working
capacity the agreement between the two tests was good
regarding heart rates at maximum work load. In the 8
patients with a low working capacity the results of the two
tests differed widely. In the orthostatic test the difference in
increase of the heart rate after standing was less than 5
beatslmin in 3 patients in one of the tests, and in 2 patients
in both tests. The exercise test in patients with atrial fibrilla-
tion must be evaluated with caution and has an acceptable
reproducibility only in patients with fairly good functional
capacity.

INTRODUCTION

We have previously reported on exercise tests in
patients with atrial fibrillation (2, 3).

The average increase in ventricular rates at
gradually increased exercise loads was often found
to be close to a linear repationship. In particular,
this was the case in patients with a relatively well-
maintained functional capacity of the heart. How-
ever, in patients with more severely impaired heart
function there was a tendency of a slightly more
pronounced increase in the heart rate from resting
value to especially that of the first load. In the
literature different results have been presented on
this problem (4, 7, 12).

In another study we have investigated the influ-
ence of different dose rates of digitalis on the heart
rate and on the working capacity in patients with
atrial fibrillation during exercise tests (3). By in-
creasing the digitalis dosage the heart rate at the
same load decreased.

The aim of the present study was to test the
reproducibility of orthostatic and exercise tests in

patients with atrial fibrillation. In such an investiga-
tion it is of critical importance that the cardiac state
be stable and unchanged at, and in the interval
between the two exercise tests.

MATERIAL AND METHODS
In 15 patients with atrial fibrillation exercise tests were
done twice in each patient. The composition of the patient
group is shown in Table I. The group is dominated by
patients with valvular heart disease, often referred to this
hospital for consideration for cardiac surgery. The exer-
cise test was a part of this evaluation. Most patients had
fairly advanced valvular disease. The repeated test was
only performed with the patient’s consent since this test
was made mainly for scientific reasons.

This second exercise test was performed only in those
patients fulfilling certain criteria to ensure a clinically sta-
ble condition of the patient between the time of the two
tests. These criteria were:

(A) No change had occurred in clinical findings o r in
the patient’s history of his capacity between the two tests.
The patient was not in heart failure.

(B) The time interval between the two tests had not
exceeded 3 months and i t had frequently been much less.
Seven of the cases had the tests less than 2 weeks apart.

(C) The resting ECGs were identical on both occasions.
In 3 patients with more than 1 month between the tests
chest X-rays were performed before the two exercise
tests and were found to be identical.

(D) The patient had the same drugs in identical dosages
on the two occasions. Serum potassium and other electro-
lytes were within normal limits at both tests.

(E) N o other disease had occurred during the interval
between the two tests.

The orthostatic test preceding the exercise test was
performed and described in more detail by Sandberg (9).
According to the recurrent nomenclature the test was
considered ‘negative’ when the heart rate increased less
than 20 beatslmin and ‘positive’ if between 20 and 29.

The exercise test was performed according to the
graded ‘steady state’ principle (10, 13). The exercise test
was performed using a bicycle ergometer (5, 6) with the
method and calculations having previously been described
(2). The heart rate was determined a t rest, after 8 min

Uosciki J M i 4 Sci 82

28 H . Aberg et al.

Table I. Composition of the case material

Diagnosis

No. of patients Age Heart volume cc/m2 BSA

Male Female Total Mean Range Mean Range

Mitral stenosis 4 3 7 42 28-57 745 510-1 060

(650) disease 1 1 2 (49) - -

valve disease 2 3 5 48 39-62 730 590-1 010
Unknown 1 1 (36) (550) -

Combined mitral

Mitral +aortic

Total 8 7 15 44.8 28-62 714 510-1 060

standing, and after 2, 4 and 6 min a t each load as the
average of 25 consecutive heart intervals. The test was
supervised by a physician at all times.

The results of the exercise test was expressed in differ-
ent ways: as the work load at a heart rate of ll0/min
(W,,,), as the highest work load which according to clini-
cal judgment of the patient’s symptoms and signs was
permissible and possible to perform for 6 min (W,,,), and
as the ability to reach ‘steady state’ as defined below. W,,,
was calculated by slight interpolation or extrapolation,
assuming a linear relationship between heart rate and
work load within the small interval in question. In patients
with atrial fibrillation and very low working capacity this
is less precise (2). A ‘steady state’ was judged to be
obtained when the difference between the heart rates a t 2,
4 and 6 min on the same load was 10 beats/min or less.

RESULTS AND COMMENTS

The resting heart rates were fairly constant on the
two occasions with few exceptions. In 13 of IS pa-
tients the difference in resting heart rates were 10
beats/min or less and in 6 of 15 only 5 beatslmin or
less. The resting values (the lowest of the 2) were
below 70 beats/min in 9 patients and above 90 (the
highest of the 2) in 3 patients. These rates were
rather high, considering that all patients except one
was on maintenance dose digitalis.

With regard to the orthostatic test there was a
difference between the two tests in 3 patients (from
‘positive’ t o ‘negative’ or vice versa). Only in one
was the difference greater. This patient had on the
first occasion an increase of the heart rate of 26
beatslmin after 8 min standing (‘positive’ reaction)
and on the second test an increase of 6 beatslmin
(‘negative’ reaction). The other 2 had 18 and 16 on
the first and 23 and 25 on the second test, respec-
tively. In S cases the difference was 5 beatslmin or
less between the two tests of orthostatic reaction.

The working capacity, measured as Wllo, was
generally low, with a few exceptions (Table II). The
mean W,,, was 168 (range 18439) and 191 (range

Upsulu J Med Sci 82

23497) kpm/min at the two exercise tests. This was
expected, as many patients had fairly advanced
heart disease. In 8 patients with a W,,, less than 100
kpm/min in one or both of the exercise tests there
were 4 patients whose difference between the two
tests did not exceed 1/3 of the lowest value. In one
the difference was about 40%. In the remaining 3 in
this patient group with low working capacity the
difference was as much as double or more the low
W,,, in each pair. In the group with a working
capacity W,,, above 110 kpmlmin, consisting of 7
patients, the agreement between the two exercise
tests was slightly better. In 3 patients the results

Table 11. Results of exercise tests
A is the first exercise test and B the second one. Defini-
tions are given in Methods

Patients

w,,, W,,,
(kpm/min) (kpmlmin)

A B A B

1 . A . M .
2. R . B .
3 . V . R .
4 . K . B .
5 . A . K .
6. V. P.
7. M. S.
8. N . E .
9. G . H .

10. E. 0.
1 1 . S. K.
12. E . W .
13. E. 0.
14. G. S .
15. S . H .
Mean
S.D.

18 23
50 36

177 169
88 1 I6

180 82
23 30

308 391
167 86
206 248
75 260

3 I7 497
I92 139
37 47

239 376
439 367
168 191
123 155

150
1 50
200
200
200
250
250
300
400
400
400“
450
500
500
700
337
158

1 50
1 50
200
200
200
250
250
300
400
400
600
500
500
500
700
3.53
175

a The patient showed aberrant ventricular ECG-
complexes. The first test was then interrupted. In spite of
the same occurrence of aberration at the second test the
exercise was allowed to continue.

Reproducibility of exercise in atrial fibrillation 29

Table 111. Heart rate resting, standing and during exercise
A is the first exercise test. B-A is the difference between the second and the first test. W , is the first load and Wz is the
second load. Patients 1-8 have the lowest W,,, and patients 9-15 have the highest W,,,. Definitions in Methods

Patients 1-8 Patients 9-15 Total

Mean S.D. Mean S . D . Mean S.D.

Resting
A
B -A

A
B-A

W,,,
A
B-A

Standing

W ,

75.0
2.1

10.4
-1.6

104.8
4.9

125. I
5.8

129.4
10.3

15.3
4.0

9.9
6.7

19.0
10.9

25.4
12.0

22.1
11.6

77.4
-7.3

17.3
-0.2

107.6
-6.8

134.0
-5.4

154.2
-4.3

9.8
6.7

5.6
3.9

23.1
8.4

26.1
9.6

18.1
9.6

76.1
-2.2

13.6
-0.9

105.8
0.4

128.5
1.5

137.5
3.1

12.6
4.3

8.7
5.4

19.8
10.0

25 .o
11.0

22.4
11.3

were almost identical, 2 had a difference of about
1/3 and the remaining 2 had about 50% increase
from the lowest value in the compared pair.

W,,, values were identical in 13 of the 15 pa-
tients. In one patient, on the second occasion, the
maximal load was chosen as 500, due to a mistake,
instead of as planned 450 kpmlmin, while at lower
loads that were the same there was a good agree-
ment between the two tests. In another patient,
there were similar heart rates at 200 and 400
kpmlmin but on the first exercise test the in-
vestigator finished the test due to aberrant veptricu-
lar ECG-complexes. The second test was contiwed
in spite of a similar aberration.

In 3 patients with a W,,, of 500 kpmlmin and
more the highest attained heart rates were almost
identical at the two tests, 150 and 146 in one, 171
and 171 in another and 127 and 123 in the third,
respectively. In 2 patients who had a W,,, of 400,
one had heart rates of 141 and 134 but the other
patient had 163 and 137. Out of the 8 patients with a
low W,,,, 2 had shown a good agreement between
the heart rates at the two tests. In Table I11 the dif-
ferences between the patients with a low and those
with a higher W,,, are seen. It is apparent from the
Table that the S.D. of the differences between the
two exercise tests is smaller in the group with a
better W,,,. The only exception is at rest where the
S.D. was higher in the group with a higher W,,,.

A ‘steady state’ was achieved on both occasions

in only 2 patients. In 9 patients a ‘steady state’
condition at the highest load was obtained during
one exercise test but not the other. Six of these
patients obtained ‘steady state’ conditions on the
second exercise test.

DISCUSSION

In many patients the case history is sufficient for
a clinically appropriate judgment of the working
capacity. However, i t is sometimes difficult to get a
reliable description from the patient and also i t is
often necessary to obtain a more objective and pre-
cise measurement on the functional capacity.

This is for example necessary to evaluate results
after medication, surgery, or in an attempt to find
an optimal time for surgical intervention etc. The
need of a reliable method in this respect is obvious.
Such a method is of course to a critical degree
dependent upon a good reproducibility of the pro-
cedure. The ideal method should be well
standardized, easy to repeat and the results should
not be influenced by the training effect of repetition.
The method used in the present study probably
fulfils these requirements. A training effect of any
significance is not probable considering the time
lapse between the two tests and the fact that the test
was repeated only once.

A most important factor with regard to the
purpose of this study, i.e. to evaluate the repro-

30 H . Aberg et al.

ducibility of the exercise test is the stability of the
patients with regard to their cardiac state. Most
patients had got their atrial fibrillation in the course
of a progressive heart disease and it was difficult to
ascertain that the patients were stable enough for
the purpose of this investigation.

I n spite of a careful selection of patients, the
results of the exercise tests in the patients with a
low working capacity indicate either that such
stable conditions were not obtained or that the ar-
rhythmia was highly variable in itself. In the group
of patients with a better heart function there was a
relatively good reproducibility of the orthostatic
test reaction as well as the exercise test.

The ventricular heart rate in atrial fibrillation is
dependent upon the conditions of the atrioventricu-
lar junction (7, 1 I ) . It has been shown that the rate
of the atrial activity is not changed during exercise
in patients with atrial fibrillation (1). In spite of the
irregular atrial activity the atrial rate is so rapid that
there is always an impulse on hand for propagation
through the atrioventricular junction. Therefore,
the ventricular response in atrial fibrillation should
be dependent upon the function of the atrioventricu-
lar node. Thus, we must consider influences from
pharmacologic agents and the autonomous nervous
system as well as from the heart disease in itself. I n
a previous paper we have shown that a rather high
digitalis dosage is favourable for patients with atrial
fibrillation (3). This has also been studied by Red-
fors (8).

In this study, however, there was no difference in
drug therapy or electrolyte balance between the two
tests. The alteration in atrioventricular function,
particularly in those individuals with low working
capacity, should therefore be caused either by vari-
ation in autonomic tone or by changes in the under-
lying heart disease.

In conclusion, the reproducibility of exercise
tests was acceptable in patients with a good work-
ing capacity. On the other hand, in the group with
more severely impaired heart function, the repro-
ducibility was poor in spite of digitalization.

REFERENCES
1. Aberg, H . & Furberg, B.: Atrial activity during exer-

cise in patients with atrial flutter or atrial fibrillation.
Ups J Med Sci80: 20, 1975.

2. Aberg, H . , Strom, G. & W e r n e r , I.: Heart rate during
exercise in patients with atrial fibrillation. Acta Med
Scand 191; 3 15, 1972.

3 .

- The effect of digitalis on the heart rate during
exercise in patients with atrial fibrillation. Acta Med
Scand 191:441, 1972.
Holmgren, A , , Jonsson, B . , Linderholm, H . ,
Sjostrand, T. & Strom, G . : Physical working capacity
in cases of mitral valvular disease in relation to heart
volume, total amount of hemoglobin and stroke vol-
ume. Acta Med Scand 167: 99, 1958.
Holmgren, A. & Mattsson, K . H . : A new ergometer
with constant load at varying pedalling rate. Scand J
Clin Lab Invest6: 137, 1954.
Holmgren, A . & Strandell, T.: On the use of
chesthead leads for recording of electrocardiogram
during exercise. Acta Med Scand 169: 57, 1961.
Know, J. A . C.: The heart rate with exercise in pa-
tients with auricular fibrillation. Br Heart J ll: 119,
I 949.
Redfors, A , : Digoxinbehandling vid formaksflirnmer.
Relationen mellan dos, plasmakoncentration och ef-
fekt. Akademisk avhandline., Lund 1971.

9. Sandberg, L.: Studies on electrocardiographic
changes during exercise tests. Acta Med Scand
[Suppl.] 36.5, 1961.

10. Sjostrand, T.: Changes in the respiratory organs of
workmen at an ore smelting works. Acta Med Scand
[Suppl.] 196: 687, 1947.

1 1 . Soderstrom, N.: What is the reason for the ventricular
arrhythmia in cases of auricular fibrillation? Am
Heart J40:212, 1950.

12. Varnauskas, E., Cramkr, G . , Malmcrona, R., Dahl,
L.-E., Nystrom, B., Wassen, A. & Werko, L.: Res-
toration of normal sinus rhythm in patients with
mitral-valve disease and atrial fibrillation. Nord Med
62: 1109, 1959.

13. Wahlund, H.: Determination of the physical working
capacity. Acta Med Scand [Suppl.]215, 1948.

Received September 2, 1976

Address for reprints:

H a n s Aberg, M.D.
Department of Internal Medicine
University Hospital
S-750 14 Uppsala 14
Sweden

Upsula .I M e d Sci 82