Vol.1, No.3, 188-191 (2009)
doi:10.4236/health.2009.13031
SciRes
Copyright © 2009 Openly accessible at http://www.scirp.org/journal/HEALTH/
Health
No relationship between the incidence of
fentanyl-induced cough and smoking
Boris Pokhis, Hans-Bernd Hopf
Department of Anesthesia and Perioperative Medicine, Asklepios Klinik Langen, Academic teaching hospital of the Johann Wolfgang
Goethe University Frankfurt am Main, Langen, Germany; corresponding author: h-b.hopf@asklepios.com
Received 17 September 2009; revised 22 September 2009; 23 September 2009.
ABSTRACT
The purpose of the study was to investigate
whether or not the incidence of cough after intra-
venous fentanyl depends on the patient’s smoking
state and the speed of injection. 530 ASA class I-III
patients free of bronchial hyperreactivity and res-
piratory tract infection undergoing general anes-
thesia for elective surgery were randomized to 1.5
g.kg-1 fentanyl injected over 2, 5 or 10 sec or pla-
cebo via a peripheral intravenous cannula. The
endpoint was cough within 5 min after completion
of injection. Statistical evaluation was performed
by factorial ANOVA and chi-square-test. Assuming
around 25% smokers in our patient population
calculated patient sample size was 340 per group.
The study was terminated for futility after enrol-
ment of 530 patients since an interims analysis
yielded an incidence of cough of 2 % both in
smokers (n=174) and nonsmokers (n=356, p=
0,970), which was unrelated to the speed of injec-
tion and not different from placebo.
Keywords: Cough, fentanyl induced; smoking;
fentanyl induced cough; cough; fentanyl;
speed of injection
1. INTRODUCTION
Cough after intravenous fentanyl for induction of anes-
thesia has been overestimated for more than 15 years. The
question of fentanyl-induced cough is clinically of con-
siderable importance since several prophylactic measures
haven been proposed to suppress cough after injection of
fentanyl (2-7) all with specific possible side effects. We
have shown in ASA I-III nonsmoking European patients
that the incidence of cough after 1.5 g.kg-1 fentanyl is
around 4% and not related to the speed of injection [1],
much lower than previously assumed [2-9].
Fentanyl induced cough is also believed to be much
lower in smokers compared to nonsmokers (3% vs 13%)
with the incidence depending on the speed of injection [8].
We hypothesised, based on our own data in nonsmoking
European patients that there would be no clinically rele-
vant difference between smokers and nonsmokers in fent-
anyl-induced cough at least in a European population.
Accordingly, to evaluate the influence of smoking on the
incidence of fentanyl induced cough we performed a
single blinded, randomized controlled study with pe-
ripheral intravenous injection of 1.5 g.kg-1 bw fentanyl. In
addition we hypothesizedd that in smokers the speed of
injection is also not related to the incidence of cough.
2. METHODS
With institutional and ethical committee approval and after
having obtained informed written consent 542 ASA class
I-III patients without history of bronchial hyperreactivity
and free of any upper respiratory tract infections scheduled
for elective general, orthopedical, or gynecological surgery
with general anesthesia were randomised (computer gen-
erated list, permuted blocks of 12). The study was per-
formed between Oktober 1th 2007 and May 31th 2008.
Patients received either 1.5 g.kg-1 bw. fentanyl injected
over 2, 5 or 10 seconds or saline placebo (NaCl 0.9%)
injected over 2 sec. The study was carried out in a single
blinded design i.e. patients were blinded concerning drug
and speed of injection. Patient allocation was performed
consecutively in the anesthesia pre-evaluation ambulance
by the attending anaesthesiologist according to numbers
generated by the computer generated list.
During the preparation of our previous study all anes-
thesia nurses (n=7) had been trained by means of a stop
watch in delivering the three predefined injection velocities.
After placement of the routine monitors (noninvasive
blood pressure, ecg, pulse oximetry) the anesthesia nurse in
charge injected the assigned drug within the assigned time
controlled by a stop watch. After completion of injection
patients were observed for 5 minutes. Througout the 5 min
study period all patients were preoxygenated (6 l.min-1 flow
*Presented in part at the German anesthesia annual congress, Leipzig,
May 9th, 2009.
B. Pokhis et al. / HEALTH 1 (2009) 188-191
SciRes Copyright © 2009 Openly accessible at http://www.scirp.org/journal/HEALTH/
189
189
O2, semi-closed system with the expiration valve of the
anesthesia circuit being set to 5 cm of water). Drugs were
always injected at room temperature via a periphereal intra-
venous cannula at the back of the right or left hand in a
running infusion of Ringer’s lactate.
Primary endpoint was cough within 5 min after com-
pletion of injection. Cough was defined as any tussive
reaction not consistent with normal quite breathing. Pri-
mary outcome was assessed by the attending anaesthesi-
ologist. Secondary endpoints were nausea and vomiting or
any other complaints of the patients within 5 min after
completion of injection.
Basic assumptions were set as in our previous study: a
5% incidence of fentanyl-induced cough in the fentanyl
and of 0% in the placebo group, an alpha error of 0.01 and a
ßeta error of 0,05 resulting in 340 patients per group [10].
Assuming a portion of 25% smokers in our local popula-
tion an interim analysis was planned after enrolment of 500
patients, when in all likelihood enough (more than 100)
smokers were included into the study protocol to allow
statistically meaningful results.
Statistical evaluation was performed with an analysis of
variance for continuous and the chi-square-test for cate-
gorical variables. We compared the incidence of cough
between smokers and nonsmokers in the whole group and
the incidence of cough between fentanyl and placebo with-
in the smoking and the non-smoking groups, respectively.
Our null hypothesis was that there were no significant
differences in the incidence of cough between smokers and
nonsmokers and between the fentanyl groups and saline
placebo in the smoking and non-smoking group, respec-
tively. After adjusting for multiple testing (bonferroni`s
method) the null hypothesis was rejected and statistical
significance assumed with a p value less than 0.001.
3. RESULTS
The flow chart of the patients of our study is shown in Fig-
ure 1. Overall, from 542 patients enrolled, 530 received the
allocated medication. Eleven patients did not present
Allocation
Analyzed n=133
Excluded from analysis: n=0
Lost to follow-up: n=0
Discontinued intervention: n= 0
Allocated to Fentanyl 2 sec: n= 136
Received Fentanyl 2 sec: n= 133
Did not receive allocated
intervention:
n= 3 not presented on scheduled
operation day
Excluded: n=0
Not meeting inclusion criteria: n=0
Refused to participate: n=0
Other reasons: n=0
Lost to follow-up n=0
Discontinued intervention n=0
Allocated to Fentanyl 5 sec: n=135
Received allocated intervention:n=
133
Did not receive allocated
intervention: n= 2 not presented on
scheduled operation day
Allocated to Fentanyl 10 sec: n= 136
Received allocated intervention: n=
132
Did not receive allocated
intervention: n= 4 not presented on
scheduled operation day
Allocated to Saline Placebo: n= 135
Received allocated intervention:
n=133
Did not receive allocated
intervention: n= 2 not presented on
scheduled operation day
Lost to follow-up: n=0
Discontinued intervention n=0
Lost to follow-up: n=0
Discontinued intervention n=1
Analyzed n=133
Excluded from analysis: n=0
Analyzed n=131
Excluded from analysis:
n=1
Analyzed n=133
Excluded from analysis:
n=0
Assessed for eligibility (n=542)
Enrollment
Randomization
Follow-Up
Analysis
Figure 1. The consort E-flowchart.
B. Pokhis et al. / HEALTH 1 (2009) 188-191
SciRes Copyright © 2009 Openly accessible at http://www.scirp.org/journal/HEALTH/
190
at the operation day scheduled, equally distributed in the
four groups. In one patient in the fentanyl 10 sec group the
syringe with the study medication was dislocated during
the injection. Therefore the patient was excluded from
evaluation. Thus, enrolment, allocation and follow up of
the patients were highly conformable to the study proto-
col.
The study was terminated for futility after enrolment of
530 patients since the planned interims analysis yielded
an incidence of cough in 4/174 patients (=2%) in the
smoking and of 8/356 patients (=2%) in the non-smoking
group (p=0.970), Table 1 shows the results of the three
injection velocities in the smoking, non-smoking and overall
patient groups compared to placebo respectively. There were
no differences in fentanyl induced cough within the three
fentanyl groups of different injection velocities neither
when comparing between the fentanyl groups nor when
comparing the fentanyl groups to placebo.
Only one patient in the fentanyl 2 sec and one patient in
the 5 sec group (in the smoker group) reported nausea
during the study period. No patient suffered from vomiting
during the study period.
Demographic data (age, height, weight, distribution of
sex) and ASA class did not differ between groups.
4. DISCUSSION
The main finding of our study is that the incidence of cough
after 1.5 g.kg-1 fentanyl is 2% both in smokers and non-
smokers. Within the smoking and non-smoking group it
was also not related to the speed of injection and not differ-
ent from saline placebo. Thus, this study confirmed our
previous findings of a clinically irrelevant incidence of fen-
tanyl induced cough in European people also in smokers.
Our results emerged when confounding variables such
as patient population, different doses of fentanyl or
speed or site of injection either were strictly controlled
or eliminated. In addition, our study is the first, which
compared in smokers different injection velocities not
only between the fentanyl groups but also to saline pla-
cebo.
As in our previous study in non-smoking patients [1]
we have choosen the 1.5 g.kg-1 dose of fentanyl, because
it is the routine used induction dose in > 90% of our pa-
tients. While even 7 g.kg-1 fentanyl injected in a pe-
riphereal vein over 1 second evoked cough in only 1/37
patients in an European population [8], two studies in an
asiatic/chinese population testing an identical dose of 1.5
g.kg-1 fentanyl found much higher incidences of up to 28
% [6,9], suggesting a race dependence.
Regardless of the smoking status, injection velocity
did not influence the incidence of cough and was con-
siderably lower than previously reported [6]. Since,
compared to our study, the dose of fentanyl as well as the
site of injection were identical, other reasons must be
responsible. First, in the study of Lin et al. [6] the speed
of injection (2, 15 and 30 seconds) was considerably
slower than in our study (2, 5 and 10 seconds). Assuming
any relation between the incidence of cough and the
speed of injection a slower injection velocity should re-
sult in a lower incidence of cough. However, compared
to our results the incidence of cough in the study by Lin
was greater with slower injection velocities, excluding
any possible relationship. Second, the study of Lin et al
[6] and ours differed in the substantially longer post-
injection observation period in our study (30 seconds in
the study by Lin et al. vs. 5 minutes in our study proto-
col). Since a longer observation period in all likelihood
should result in a higher rather than a lower incidence of
cough by chance alone this also can be excluded as an
explanation for the higher incidence of cough in the
study by Lin et al. Thus racial differences of the subjects
studied (Chinese vs. European people) seemed to be the
most likely explanation for the different incidences of
Table 1. Incidence of cough after intravenous injection of fentanyl or placebo( n=530).
Fentanyl 2 sec (n=133)Fentanyl 5 sec (n=133)Fentanyl 10 sec
(n=131) Placebo (n=133)
Cough yes/no
( %)
Cough yes/no
( %)
Cough yes/no
( %)
Cough yes/no
( %)
Smoker (n=174) 1/37 1/44 1/46 1/43
(2,6%) (2,2%) (2,1) (2 ,2%)
p=0,999
Nonsmoker (n=356) 3/92 2/86 1/84 2/86
(3,1 %) (2,2%) (1,0%) (2,2 %)
p=0,848
Total (n=530) 4/129 3/130 2/129 3/130
(3,0 %) (2,2%) (1,5%) (2,2 %)
p=0,884
Values shown are numbers. Values in parenthesis represent percentage of patients with cough. P-values refer to the chi-square-test.
B. Pokhis et al. / HEALTH 1 (2009) 188-191
SciRes Copyright © 2009 Openly accessible at http://www.scirp.org/journal/HEALTH/
191
cough between the study by Lin et al. compared to our
study. In fact, in an asiatic or chinese population
[2,3,6,7,9] intravenous injection of fentanyl in doses ran-
ging from 2-5 g.kg-1 given within 1-5 seconds evoked
cough in 28-65% in contrast to 3% incidence after 7 5
g.kg-1 given within 1 second in an European population
[8].
That in our study cough was also evoked after saline
placebo injection in all likelihood is the result of our
considerably longer post-injection observation period (5
minutes) in conjunction with preoxygenation by face
mask, both increasing the chance for cough either by
time alone or by inconvenience of the mask regardless of
the drug injected. Consistent with this assumption, in the
placebo groups cough occurred in 2 patients after 1,5 and
3 minutes in the nonsmokers and after 4 minutes in one
smoker, respectively, in contrast to the fast (< 1 minute)
response after intravenous fentanyl.
In summary, in this study in an European population the
incidence of cough after intravenous injection of fentanyl
was 1) not different between smokers and nonsmokers, 2)
unrelated to the speed of injection within the smoking and
non-smoking group, and 3) not different from saline pla-
cebo. Therefore in a routine clinical setting fentanyl-in-
dued cough is a rare event without any need for prophy-
laxis.
REFERENCES
[1] U. Schäpermeier, H. B. Hopf, (2008) Fentanyl-induced
cough does not depend on injection speed: A randomised
study. Acta Anaesthesiol Scand, 52, 1071-5.
[2] P. W. Lui, C. H. Hsing, Y. C. Chu, (1996) Terbutaline
inhalation suppresses fentanyl-induced coughing. Can J
Anaesth, 43, 1216-9.
[3] A. Agarwal, A. Azim, S. Ambesh, et al., (2003) Salbuta-
mol, beclomethasone or sodium chromoglycate suppress
coughing induced by iv fentanyl. Can J Anaesth, 50,
297-0.
[4] C. S. Lin, W. Z. Sun, W. H. Chan, C. J. Lin, H. M. Yeh, M.
S. Mok, (2004) Intravenous lidocaine and ephedrine, but
not propofol, suppress fentanyl induced cough. Can J
Anaesth, 51, 654-9.
[5] C. K. Pandey, M. Raza, R. Ranjan, et al., (2004) Intrave-
nous lidocaine suppresses fentanyl-induced coughing: a
double-blind, prospective, randomized placebo-controlled
study. Anesth Analg, 99, 1696-8.
[6] J. A. Lin, C. C. Yeh, M. S. Lee, C. T. Wu, S. L. Lin, C. S.
Wong, (2005) Prolonged injection time and light smoking
decrease the incidence of fentanyl-induced cough. Anesth
Analg, 101, 670-4.
[7] H. C. Horng, C. S. Wong, K. N. Hsiao, B. K. Huh, C. P.
Kuo, C. H. Cherng, C. T. Wu, (2007) Pre-medication with
intravenous clonidine suppresses fentanyl-induced cough.
Acta Anaesthesiol Scand, 51, 862-865.
[8] H, Böhrer, F. Fleischer, P. Werning, (1990) Tussive effect
of a fentanyl bolus administered through a central venous
catheter. Anaesthesia, 45, 18-21.
[9] W. T. Phua, B. T. Teh, W. Jong, T. L. Lee, W. A. Tweed,
(1991) Tussive effect of a fentanyl bolus. Can J Anaesth,
38, 330-4.
[10] S. J. Pocock, (1998) Clinical trials: A practical aproach.
Chichester New York Brisbanae Toronto Singapore: John
Wiley & Sons, Chapter 9, 123-141.