CARDIOVASCULAR JOURNAL OF AFRICA • Vol 21, No 2, March/April 2010
AFRICA
101
critical to survival once beta-blockers are withdrawn in the peri-
operative period.
1
An increase in the mean heart rate of
≥
six
beats per minute following withdrawal of chronic beta-blockade
was found to be independently associated with peri-operative
mortality in vascular surgical patients. The positive likelihood
ratio (LR) is 3 and the negative LR is 0.2. For a test to be clini-
cally useful, it should have a positive LR exceeding 5 to 10 and a
negative LR of less than 0.2.
9,10
Based on the heart rate threshold
identified in this study and the associated LR, we can conclude
that patients who have a heart rate increase of less than six beats
per minute following withdrawal of chronic beta-blockade are
likely to survive, as the LR is within the clinically useful range.
When the heart rate increases by at least six beats per minute,
then the discrimination is not as accurate.
Therefore although this heart rate increase is an independent
predictor of mortality when chronic beta-blockade is withdrawn,
it is still clinically inaccurate, reflecting the uncertainty associ-
ated with this predictor. This poor positive discrimination is not
an unusual problem with peri-operative risk prediction.
9
Indeed,
none of the positive results of the current special investigations
for vascular surgery appear to be accurate enough for positive
discrimination.
11
The poor positive discrimination of this heart rate thresh-
old should however not be considered an important limitation,
especially when one considers how this heart rate threshold may
impact on peri-operative anaesthetic practice. When patients on
chronic beta-blockade present for surgery, keeping the heart rate
below the ischaemic threshold is a central tenant. This target
could be identified by determining the ischaemic threshold from
pre-operative Holter monitoring or by arbitrarily selecting a heart
rate threshold.
1
Certainly, with a negative LR of 0.2, the heart
rate threshold presented in this article is clinically useful. If an
anaesthetist could successfully control the heart rate below this
threshold, then it would probably be unnecessary to conduct a
pre-operative Holter in order to determine an ischaemic thresh-
old.
However, what should we do with patients in whom chronic
beta-blockade is withdrawn and who have an increase in heart
rate
≥
six beats per minute? As we cannot accurately predict
which patients will die using this threshold, but as we do know
that this group is at a significantly increased risk of in-hospital
mortality, we could use this threshold as an ‘early warning
system’. It would be reasonable to attempt to re-institute beta-
blockade as soon as possible if not contra-indicated. Transfer of
the patient to a high-care facility, serial monitoring of troponins,
repeat ECG
12
and an attempt to institute other negatively chrono-
tropic therapy if beta-blockade is contra-indicated should all be
considered.
Although withdrawal of chronic beta-blockade is clearly unde-
sirable,
1-3
it does occur, either inadvertently, or as a considered
decision based on perceived contra-indications or peri-operative
complications considered to be related to beta-blocker adminis-
tration. The findings of this study may be useful in identifying
patients at risk and initiating management and therapy that may
decrease the high associated morbidity and mortality.
There were limitations to this study. One-third of the patients
who were withdrawn from beta-blockade were excluded from the
study because of missing data. It is possible that these patients
could have had a meaningful impact on the statistical analysis. In
over 50% of the patients who had beta-blockers withdrawn and
died, there was no obvious reason for the withdrawal (Table 3).
Clearly, such a high withdrawal rate with no obvious indication
needs to be studied and the reasons for withdrawal elicited, as
omission of chronic beta-blockade is potentially life threaten-
ing.
It is important to note that even in chronically beta-blocked
patients, there is a significant circadian rhythm to the mean heart
rate with the nocturnal heart rate being lower than the daily heart
rate.
13
The ischaemic threshold is significantly decreased in the
early morning.
14
It is likely that a number of factors contribute
to increased myocardial oxygen demand in the morning, includ-
ing heart rate, blood pressure, autonomic and humeral physi-
ological changes.
15
The use of a mean daily heart rate, as used in
this study, has limitations due to the variation of the ischaemic
threshold through the course of a day. It is likely that the optimal
cut-off point for the change in heart rate is lower than six beats
per minute in the morning and possibly higher in the evening.
This will, however, only be adequately addressed with a peri-
operative study using continuous Holter monitoring.
We also did not have data on the dosage of chronic beta-
blockade that all the patients in this study received, which
may be an important determinant of the ischaemic threshold.
16
Higher doses are associated with a lower ischaemic threshold,
making it possible that an increase in heart rate of
≥
six beats per
minute may be too high for patients on a high dose of chronic
beta-blockade. This then is another area that requires further
investigation, particularly if a heart rate-based cut-off point is
to be used.
Furthermore, there were statistical limitations to this study.
The large confidence intervals for the OR of mortality asso-
ciated with an increase in postoperative heart rate (95% CI:
1.7–110) are of limited clinical utility. There are two possible
explanations for the large confidence interval: firstly, the sample
size was small and secondly, it is possible that the retrospective
data collection in this study may have increased the variability
of the heart rate data collected. A small sample size results in
less precision around an estimate of risk (and hence a larger
confidence interval) and increased variability within a group
decreases the reliability of the findings, and again, increases the
width of the confidence intervals.
17
It is important to note that
despite the wide confidence intervals in this study, the postopera-
tive increase in heart rate was still significantly associated with
mortality. It appears therefore that this was an important observa-
tion, although it is difficult to precisely quantify its risk.
Is it clinically plausible that a heart rate change of only six
beats per minute can explain an associated increased mortality?
It probably is, for a number of reasons. Firstly, a Holter study
of patients with stable coronary artery disease patients receiv-
ing beta-blocker therapy for two weeks has shown that silent
myocardial ischaemia is evident with an increase in heart rate of
as little as 12.3 (
±
1.4) beats per minute from the mean resting
heart rate, immediately prior to the onset of myocardial ischae-
mia.
18
Secondly, beta-blockade has been shown to significantly
decrease the ischaemic threshold,
16
and that increasing doses of
beta-blocker are associated with further significant reduction
in the ischaemic threshold.
16
Thirdly, the ischaemic threshold is
associated with the duration of the heart rate increase, where a
lower ischaemic threshold is evident if the heart rate increase
is prolonged.
19
Furthermore, there is an increased frequency of
myocardial ischaemia at lower heart rates in patients on beta-
