Clopidogrel versus low-dose aspirin as risk factors for
PRELIMINARY COMMUNICATIONS
epistaxis
Rainsbury, J.W. & Molony, N.C.
ENT Department, Russells Hall Hospital, Dudley, West Midlands, UK
Accepted for publication 7 March 2009
Clin. Otolaryngol. 2009, 34, 232–235
Objectives: To quantify the relative risk of epistaxis for
patients taking low-dose aspirin or clopidogrel compared
to patients taking neither drug.
Design: Case-control study.
Setting: Primary care.
Participants: 10,241 patients from three GP practices in
the West Midlands.
Main outcome measures: Epistaxis resulting in presenta-
tion to the GP, attendance at Accident & Emergency, or
referral to ENT outpatients.
Results: There was a significant difference in the
proportion of patients with epistaxis across the three
Aspirin and clopidogrel are antiplatelet drugs that are
commonly prescribed for secondary prevention of ischae-
mic vascular events, and are often taken by patients
attending ENT departments with epistaxis. Aspirin is
thought to be a risk factor for epistaxis.1 By virtue of its
similar mode of action, clopidogrel is assumed to be a
risk factor too, although it is not listed in the British
National Formulary as such, and there are no published
studies estimating the risk of epistaxis in patients taking
clopidogrel.2 Hypertension is a risk factor for ischaemic
vascular events and also possibly for epistaxis, although
this association is less clear.3,4 We aimed to find out
which drug carries a higher risk of troublesome nose-
bleeds, and whether hypertension is an independent risk
factor for epistaxis.
Method
Data collection
The electronic records of all patients (n = 24 126) regis-
tered at three GP practices in Dudley, West Midlands,
Correspondence: James W. Rainsbury, ENT Department, Russells Hall
Hospital, Pensnett Road, Dudley, West Midlands DY1 2HQ, UK.
e-mail: j_rainsbury@yahoo.co.uk
232 Journal compilation  2009 Blackwell Publishing Ltd • Clinical Otolaryngology 34, 232–235
groups (v2 ¼ 84.1; 2 degrees of freedom; P < 0.000001).
Relative risk of epistaxis was increased in both the aspirin
(RR ¼ 9.04; 95% CI ¼ 5.13–15.96) and clopidogrel
(RR ¼ 6.40; 95% CI ¼ 2.33–17.56) groups compared to
the no drug group. There was no increased risk of
epistaxis with aspirin compared to clopidogrel (RR ¼ 1.4;
95% CI ¼ 0.6–3.4).
Conclusion: There is an increased risk of troublesome
epistaxis in patients taking aspirin or clopidogrel. There
is no significant difference in risk of epistaxis between the
two drug groups.
were searched retrospectively over five years (June 2003–
June 2008). The practices were chosen because of proxim-
ity to our ENT department and willingness to take part.
The number of patients, age and sex distribution in each
practice was similar.
Patients taking combinations of anticoagulant or anti-
platelet drugs, children aged sixteen years or less, and
those with known bleeding disorders were excluded. The
remaining patient records (n = 10 241) were divided into
three groups: patients with repeat prescriptions for either
low-dose (75 mg) aspirin or 75 mg clopidogrel, and
patients taking neither drug. The groups were matched
for age and sex. An episode of epistaxis was defined as
a presentation to (1) the GP, (2) the Emergency
Department, or (3) ENT Outpatient clinic; the number of
episodes was recorded for each group. By hand-searching
the small number of case-notes of patients referred from
the GP to ENT Outpatients (n = 62), we attempted to
avoid counting the GP and ENT episodes twice.
The records were also searched for patients with a
diagnosis of hypertension. Hypertension was diagnosed if
the systolic blood pressure (BP) was >140 mmHg or the
diastolic BP was >90 mmHg on two or more occasions.
Patients were invited for a routine BP check by all prac-
tices once a year after the age of 45, or every 3 months if
their BP was elevated.
 2009 The Authors
 Clopidogrel versus low-dose aspirin 233

Table 1. Sample characteristics of the three study groups founding variable. By comparing the RR of epistaxis from
Drug % Male Mean age (SD) a single-variable logistic regression model (drug only)
with the risk calculated from a dual-variable logistic
No drug 49.2 65.8 (8.9) regression model (drug and hypertension), we were able
Aspirin 48.0 66.9 (13.8)
to estimate the confounding effect of hypertension. If
Clopidogrel 50.4 66.1 (13.3)
these figures differ by less than 5%, there is not a signifi-
cant amount of confounding.

Table 2. Number of patients presenting with epistaxis accord-

ing to drug group Results
Drug Epistaxis No epistaxis Total The patient characteristics of each group are shown in
Table 1. Results are shown in Table 2. There was a highly
No drug 16 7368 7384
significant difference in the proportion of patients with
Aspirin 48 2444 2492
Clopidogrel 5 360 365 epistaxis across the three groups (v2 = 84.1; 2 d.f.;
Total 69 10 172 10 241 P < 0.000001). Aspirin was largely responsible for this
result. Table 3 shows the number of patients with epi-
v2 = 84.1; d.f. = 2; P < 0.0001; Fisher exact test P < 0.0001. staxis according to drug group and BP status. There was
no significant difference between the groups when analy-
sed with v2 or Fisher exact tests (v2 = 0.35; 2 d.f.;
Table 3. Number of patients presenting with epistaxis accord- P = 0.84; Fisher exact P = 0.88). Patients in both drug
ing to drug group and blood pressure status groups had an increase RR of epistaxis relative to the no
Drug Hypertension No hypertension Total drug group. There was no evidence that hypertension was
a confounding factor (Table 4). Hypertension alone was
No drug 14 2 16
not an independent risk factor for epistaxis when anti-
Aspirin 39 9 48
Clopidogrel 4 1 5
platelet drugs were controlled for: the effect of hyperten-
Total 57 12 69 sion after controlling for aspirin was a RR [95%
Confidence Interval (CI)] of epistaxis = 1.22 (0.64–1.35);
v2 = 0.35; d.f. = 2; P = 0.84; Fisher exact test P = 0.88. and for clopidogrel was RR (95% CI) = 2.24 (0.52–9.63).
Aspirin was not associated with any greater risk of epi-
staxis than clopidogrel [RR (95% CI) = 1.4 (0.6–3.4)].
Table 4. Confounding effect of hypertension on relative risk of
epistaxis
Confounding Discussion
RR of effect of
The incidence of epistaxis is difficult to pinpoint, since
epistaxis RR of epistaxis hypertension
only a small proportion of patients present to a doctor
Comparison (95% CI) {A} (95% CI) {B} 1)[100({A} ⁄ {B})]
for treatment. It is estimated that 15% of the population
Aspirin 9.04 9.05 0.01 experience minor recurrent nosebleeds, but that only
versus (5.13–15.96) (5.13–15.96) about 10% of people suffering from epistaxis seek medi-
no drug cal attention, and only 0.5–1% of the whole group are
Clopidogrel 6.40 6.23 2.63 seen by an otolaryngologist.5
versus (2.33–17.56) (2.27–17.11) One particular group of drugs thought to increase the
no drug
risk of epistaxis are those affecting platelet function, of
RR, relative risk; CI, confidence interval; {A}, RR of epistaxis which aspirin and clopidogrel are the most commonly
allowing for drug alone; {B}, RR of epistaxis allowing for drug used. Aspirin irreversibly binds to cyclo-oxygenase-1 to
and hypertension. inhibit the production of prostaglandins, which mediate
pain and inflammation; and thromboxane A2, which
Statistical analysis
mediates platelet aggregation. Low-dose aspirin is used
for secondary prevention of thrombotic cerebrovascular
Statistical analysis was performed using v2 and Fisher or cardiovascular disease, and following by-pass surgery.
exact tests; the Relative Risk (RR) of epistaxis for each Clopidogrel irreversibly blocks P2Y12, an adenosine
drug was also calculated. Logistic regression analysis was diphosphate receptor on the platelet membrane,
performed to determine whether hypertension was a con- preventing fibrin cross-linkage and platelet adhesion to
 2009 The Authors
Journal compilation  2009 Blackwell Publishing Ltd • Clinical Otolaryngology 34, 232–235
234 J.W. Rainsbury & N.C. Molony
the endothelium. It is indicated for secondary prevention
of atherosclerotic events in patients with symptomatic
peripheral or cardiovascular disease; following ischaemic
stroke; acute myocardial infarction or acute coronary syn-
drome, in combination with low dose aspirin; or in
patients who cannot tolerate aspirin.2
Tay et al.1 looked at 326 patients hospitalised with epi-
staxis and compared them with matched controls in the
community and in hospital. They found a significant
association between aspirin intake and epistaxis
(P < 0.001), and a RR for hospital admission of 2.75 in
patients with epistaxis on aspirin. A literature search on
Medline and EMBASE found no studies that specifically
defined the risk of epistaxis with clopidogrel.
The relationship between hypertension and epistaxis is
unclear.3,4 In our study, hypertension was not found to
be an independent risk factor for epistaxis. There may
have been patients who had not yet been diagnosed with
high blood pressure included in the normotensive group,
although all three GP practices routinely screened patients
over 45 for hypertension on an annual basis, so this fig-
ure is likely to have been small.
The results of our study indicate that the risk of epi-
staxis in patients taking low-dose aspirin is no higher
than in those taking clopidogrel, and both drugs confer
an increased risk of bleeding compared to no drug.
Limitations
Self-medication & non-compliance
There will have been patients in the no drug group who
were taking low-dose aspirin, but bought it over the
counter, and patients who were prescribed an antiplatelet
drug, but did not take it. Unreported aspirin use in the
no drug group is likely to have increased the incidence of
epistaxis for these patients, reducing the perceived differ-
ences between the no drug group and each of the anti-
platelet groups; poor compliance with prescribed
antiplatelet therapy would theoretically reduce the inci-
dence of epistaxis in the treatment groups. Nonetheless,
there was still a significantly increased risk of epistaxis in
both drug groups, so the impact of these unknown quan-
tities was probably negligible.
Diagnosing hypertension
There may be missing BP data about patients who did
not attend for their regular BP check, or those below 45
who have undiagnosed hypertension. We have not classi-
fied the severity of hypertension, nor taken into account
the wide variety of anti-hypertensive medications used,
Journal compilation  2009 Blackwell Publishing Ltd • Clinical Otolaryngology 34, 232–235
nor established any temporal relationship between an ele-
vated BP reading and an episode of epistaxis. These issues
need to be considered when interpreting our data and
should be addressed in any future studies.
Counting epistaxis episodes
Despite our best efforts to avoid double-counting of ENT
and GP encounters, there may have still been a number
of duplicates, although this number is likely to be a tiny
percentage of the overall number of patients.
Small numbers
There were small numbers in some of the groups, partic-
ularly in those patients on clopidogrel who had epistaxis
(n = 5). The Fisher exact test can be used for samples
with small numbers in a similar way to the v2 test. This
still showed a highly significant difference (P < 0.0001)
between the groups, suggesting that any effect of this
small sample size on the overall results was minor.
Other variables
There are other risk factors for epistaxis that we did not
control for, such as rhinitis, nasal trauma or surgery, other
non-steroidal anti-inflammatory drugs, alcohol and
weather. Variables that are unevenly distributed between
study groups may cause the investigator to draw conclu-
sions from aggregated data that are contrary to the ‘true’
conclusion (Simpson’s paradox). It is unfortunately unfea-
sible to design a trial that controls for the infinite number
of potential variables, but future studies could attempt to
address more of the likely confounders mentioned above.
Pre-existing tendency to epistaxis
Some patients will have had a tendency to epistaxis before
being placed on an antiplatelet drug. Extrapolating from
the prevalence of epistaxis requiring medical attention,5
this may represent up to 10% of patients. We have
assumed that there were a similar proportion of patients
with a tendency for nosebleeds in the no drug group,
who balanced the problem out. Unfortunately, this may
still represent an area of significant bias.
Synopsis of key findings
There seems to be an increased risk of troublesome epi-
staxis in patients taking either aspirin or clopidogrel,
regardless of BP status. There is no significant difference
in risk of epistaxis between the two drug groups.
 2009 The Authors
 Clopidogrel versus low-dose aspirin 235

Acknowledgements
Many thanks to Peter Nightingale (Queen Elizabeth Hos-
pital, Birmingham) for his invaluable help with statistics.
Conflict of interest
None to declare.
References
1 Tay H., Evans J., McMahon A. et al. (1998) Aspirin, nonsteroidal
anti-inflammatory drugs, and epistaxis. A regional record linkage
case control study. Ann. Otol. Rhinol. Laryngol. 107, 671–674
2 Mehta D.K. (2007) British National Formulary. 54th edn. British
Medical Association and Royal Pharmaceutical Society of Great
Britain, London
3 McGarry G.W. (2008) Epistaxis. In Scott-Brown’s Otorhinolaryn-
gology Head & Neck Surgery, 7th edn, Gleeson M. (ed.),
pp. 1596–1608. Hodder Arnold, London
4 Herkner H., Laggner A.N., Mullner M. et al. (2000) Hypertension
in patients presenting with epistaxis. Ann. Emerg. Med. 35, 126–
130
5 Benninger M. & Marple B. (2004) Minor recurrent epistaxis:
prevalence and a new method for management. Otolaryngol. Head
Neck Surg. 131, 317–320

 2009 The Authors

Journal compilation  2009 Blackwell Publishing Ltd • Clinical Otolaryngology 34, 232–235