ORIGINAL ARTICLE

Risk Factors for Recurrent Spontaneous Epistaxis

Victor Abrich, MD; Annabelle Brozek, MD; Timothy R. Boyle, MD;
Po-Huang Chyou, PhD; and Steven H. Yale, MD
Abstract
Objective: To identify risk factors associated with spontaneous recurrent epistaxis.
Patients and Methods: This was a retrospective cohort study assessing patients in the Marsheld Clinic
system diagnosed as having epistaxis between January 1, 1991, and January 1, 2011. There were 461 cases
with at least 2 episodes of spontaneous epistaxis within 3 years and 912 controls with only 1 episode in the
same time frame. More than 50 potential risk factors were investigated, including demographic features,
substance use, nasal anatomical abnormalities, nasal infectious and inammatory processes, medical comorbidities, medications, and laboratory values. A Cox proportional hazards regression modeling approach was
used to calculate hazard ratios of epistaxis recurrence.
Results: Traditional risk factors for epistaxis, including nasal perforation, nasal septum deviation, rhinitis,
sinusitis, and upper respiratory tract infection, did not increase the risk of recurrence. Signicant risk factors
for recurrent epistaxis included congestive heart failure, diabetes mellitus, hypertension, and a history of
anemia. Warfarin use increased the risk of recurrence, independent of international normalized ratio.
Aspirin and clopidogrel were not found to increase the risk of recurrence. Few major adverse cardiovascular
events were observed within 30 days of the rst epistaxis event.
Conclusion: Congestive heart failure is an underappreciated risk factor for recurrent epistaxis. Hypertension and diabetes mellitus may induce atherosclerotic changes in the nasal vessels, making them friable
and more at risk for bleeding. Patients with recurrent epistaxis may also be more susceptible to developing
anemia. Physicians should promote antiplatelet and antithrombotic medication adherence despite an
increased propensity for recurrent epistaxis to prevent major adverse cardiovascular events.
2014 Mayo Foundation for Medical Education and Research

From the Department of Internal Medicine (V.A., A.B.,

S.H.Y.) and Department of
Otolaryngology (T.R.B.),
Marsheld Clinic, and
Biomedical Informatics
Research Center, Marsheld
Clinic Research Foundation
(P.-H.C.), Marsheld, WI. Dr
Abrich is now with the
Department of Cardiovascular Disease, Mayo Clinic
Arizona, Scottsdale.

1636

pistaxis is a common otolaryngologic

condition, accounting for approximately
1 in 200 emergency department visits in
the United States.1 Most episodes of epistaxis are
self-limited,2 and few patients seek medical
attention.3 However, a subset of patients with
epistaxis will experience frequent recurrent episodes.4 Many putative risk factors for single
epistaxis events have been described, including
facial injury, physical and chemical irritation,
allergic rhinitis, viral and bacterial rhinosinusitis,
nasal tumors, temperature, and humidity.5 Patients may be predisposed to epistaxis as a result
of impaired hemostasis caused by antithrombotic and antiplatelet medications, certain herbal
supplements, thrombocytopenia, uremia,6
alcohol use,7 or deciency in the production
of clotting factors in liver disease.8 Frequent
epistaxis may also be observed in patients
with inherited coagulopathies, such as von
Willebrand disease, and genetic abnormalities,
such as hereditary hemorrhagic telangiectasia
(HHT). Epistaxis can also occur as a result of

Mayo Clin Proc. 2014;89(12):1636-1643

medication exposure and is associated with

the use of antiplatelet medications for the secondary prevention of heart attack and stroke,
including aspirin and clopidogrel.
In addition to the known risk factors for
epistaxis described previously herein, other potential risk factors have been described, but
causality has not been demonstrated. Nasal
septum abnormalities have often been implicated, although their association is unclear.9
Elevated blood pressure characteristic of uncontrolled hypertension is also thought to precipitate epistaxis; however, anxiety in the
context of an epistaxis episode may explain
such ndings. Long-term, uncontrolled hypertension is also believed to predispose patients
to epistaxis through the development of atherosclerosis in vessels of the nasal mucosa.5 Postmortem studies have reported greater nasal
vascular damage marked by degenerative
brotic changes in the tunica media of individuals with epistaxis who had chronic hypertension compared with individuals with epistaxis

Mayo Clin Proc. n December 2014;89(12):1636-1643 n http://dx.doi.org/10.1016/j.mayocp.2014.09.009

www.mayoclinicproceedings.org n 2014 Mayo Foundation for Medical Education and Research

RISK FACTORS FOR RECURRENT SPONTANEOUS EPISTAXIS

alone.10 These friable vessels may increase the

severity of bleeding during an acute episode
of epistaxis.11
In most patients, epistaxis is self-limiting
and causes, at worst, anxiety and discomfort.
However, larger bleeds can lead to hemodynamic instability,2 especially in elderly patients
with coexisting cardiovascular disease.9 Complications from blood loss, including angina
and myocardial infarction, have also been
described.12 Frequent nuisance bleeding,
although not as immediately dangerous, may
also have an impact on patient quality of life
and has been linked to premature discontinuation of antiplatelet medications and increased
risk of myocardial infarction from in-stent
thrombosis.13
Although many risk factors for individual
episodes of epistaxis are known, risk factors
that predispose patients to recurrent episodes
of epistaxis have not yet been described. The
goal of this study was to identify risk factors
specic to recurrence.
PATIENTS AND METHODS
This was a retrospective cohort study assessing patients in the Marsheld Clinic system
(Marsheld, Wisconsin) diagnosed as having
epistaxis between January 1, 1991, and
January 1, 2011. Cases were dened as having at least 2 episodes of epistaxis requiring
medical care separated by a minimum of 3
months within a 36-month period, whereas
controls had only 1 episode in the same
time frame. An episode of epistaxis consisted
of a cluster of nontraumatic nosebleeds,
including the incident event, subsequent
care, and follow-up. Manual adjudication of
epistaxis episodes in the electronic medical
record led to the identication of 461 cases
of recurrent epistaxis for study inclusion. A
total of 912 manually adjudicated controls
were 2:1 frequency matched to cases. Time

to recurrence was incorporated into the

matching process as follows: the 3 years after
a controls epistaxis event were divided into
3-month intervals. A control was eligible to
be matched to a case if he or she had a clinic
visit occurring during a 3-month interval corresponding to the time elapsed from a cases
rst episode to his or her second episode,
ie, time to recurrence (Figure). Two controls
were matched to every case in this manner
for 451 cases, with the remaining 10 cases
matched to 1 control owing to a limitation
in the number of eligible controls.
Patients were eligible for study inclusion if
they were aged 18 years or older, had sought
medical attention for epistaxis in the Marsheld Clinic system during each episode, and
had been seen for follow-up. Only patients
who obtained most of their health care
through the Marsheld Clinic were included
in the study to ensure that longitudinal
follow-up data were available. Patients had
to have been seen by a primary care provider
at least once within a 6-year window surrounding the incident event (up to 3 years
before or after the event). They also had to
be seen at least once by a primary care provider for follow-up within 3 years after the
incident event. Patients were excluded if they
had a history of blunt trauma or digital trauma
to the nose, a history of nasal or sinus surgery,
a known bleeding disorder (eg, von Willebrand disease), or a genetic predisposition to
nosebleeds (eg, HHT). Data regarding demographic characteristics, substance use, treatment, comorbidities, medications, laboratory
values, and adverse outcomes were electronically and manually abstracted for each
epistaxis event. For controls, the second point
of data collection was taken from the 3-month
interval corresponding to a cases second
epistaxis episode as a substitute for epistaxis
recurrence. Additional data related to adverse

Initial event

Subsequent event

Case
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
Months

Control
0
Initial event

12

15

18

21
24
Clinic visit

27

30

33

36

FIGURE. Matching a control to a case based on time to recurrence.

Mayo Clin Proc. n December 2014;89(12):1636-1643
www.mayoclinicproceedings.org

http://dx.doi.org/10.1016/j.mayocp.2014.09.009

1637

MAYO CLINIC PROCEEDINGS

TABLE 1. Demographic and Substance Use Risk Factorsa

Variable

Casesb

Controlsb

Age (y), mean  SD

Male sex
White race
Hispanic ethnicity
Smoking
Alcohol use

67.017.2
268/461 (58.1)
402/407 (98.8)
2/401 (0.5)
191/406 (47.0)
177/383 (46.2)

P valuec

HR (95% CI)

66.717.2 1.008 (1.003-1.014)

459/912 (50.3) 1.17 (0.97-1.41)
872/881 (99.0) 0.72 (0.30-1.73)
2/860 (0.2)
1.25 (0.31-5.03)
326/828 (39.3) 1.03 (0.85-1.30)
376/772 (48.7) 0.83 (0.68-1.01)

.004
.10
.45
.75
.74
.07

HR hazard ratio.
Data are given as No./total (percentage) except where indicated otherwise. Totals may differ
since variables may not have been identiable in some subjects during manual data abstraction.
c
P values were derived using Cox proportional hazards regression modeling. The dependent
variable is time to epistaxis recurrence. The independent variables are age, male sex, white race,
Hispanic ethnicity, smoking, and alcohol use.
a

events occurring within 30 days of the rst

epistaxis event were recorded for cases and
controls and included myocardial infarction,
angina, stroke, transient ischemic attack, and
blood product infusions.
Several potential risk factors for recurrent
epistaxis were assessed by univariate statistical
analysis, including demographic features, substance use, nasal anatomical abnormalities,
nasal inammatory and infectious processes,
medical comorbidities, medications, and laboratory values. The sample size was sufcient to
provide 99% statistical power to detect a standardized effect size of 20% at a 5% signicance level with respect to the prevalence of
a given risk factor when comparing cases to
controls. The Bonferroni correction was
applied to restrict type I errors. A Cox proportional hazards regression modeling approach14
was used to derive the hazard ratio (HR), corresponding 95% CI, and P value for the association between time to epistaxis recurrence
TABLE 2. Traditional Epistaxis Risk Factors at Presentationa

Variable
Nasal perforation
Septal deviation
Septal spurs
Rhinitis (allergic or nonallergic)
Upper respiratory tract infection
Sinusitis
a

Controls
Cases
(No. [%]) (No. [%])
(n461) (n912)
8
27
8
12
7
5

(1.7)
(5.9)
(1.7)
(2.6)
(1.5)
(1.1)

12
48
12
12
14
12

(1.3)
(5.3)
(1.3)
(1.3)
(1.5)
(1.3)

P valueb

HR (95% CI)
1.98
1.07
1.98
1.38
1.01
1.15

(0.98-3.99)
(0.72-1.57)
(0.98-3.99)
(0.78-2.45)
(0.48-2.13)
(0.48-2.77)

.06
.75
.06
.27
.99
.76

HR hazard ratio.
P values were derived from Cox proportional hazards regression modeling. The dependent
variable is time to epistaxis recurrence. The independent variables are nasal perforation, septal
deviation, septal spurs, rhinitis, upper respiratory tract infection, and sinusitis.

1638

Mayo Clin Proc.

and each of the described risk factors, as

well as adverse events within 30 days of the
rst epistaxis event. This model took time to
recurrence into account for the statistical analysis of cases and absence of epistaxis recurrence, or censorship, into account for
controls. For patients taking warfarin, Fishers
exact test was used to compare the prevalence
of different international normalized ratio
(INR) ranges in cases and controls. Results
were considered statistically signicant at
P<.05. All the data analysis was performed using the commercially available statistical software package SAS version 9.2 (SAS Institute,
Inc). This study was approved by the Marsheld Clinic Institutional Review Board.
RESULTS
Patient demographic characteristics and substance use patterns for cases and controls are
presented in Table 1. A slightly higher percentage of men was noted in the case group,
but this was not statistically signicant. Age
was found to confer a slight increase in risk
of recurrent epistaxis (HR, 1.008; P.004),
but the difference in mean age between cases
and controls was less than 1 year and was unlikely to present clinical signicance. No signicant differences were observed between
the 2 groups with regard to race, ethnicity, tobacco use, or alcohol consumption.
Traditional risk factors for a single
epistaxis event were considered at the time
of initial presentation (Table 2). No signicant
differences were observed between cases and
controls with respect to anatomical ndings
on physical examination or local nasal inammatory/infectious processes. There was a slight
increase in the occurrence of rhinitis in the
case group, but this difference was not statistically signicant.
Additional potential risk factors were subsequently divided into 2 categories based on
whether they were related to cardiovascular
risk (Table 3). Of the cardiovascular risk factors
considered, congestive heart failure (CHF) was
the most strongly associated with recurrent
epistaxis (HR, 1.61; P<.001). Although associations were weaker, increased risk of epistaxis
recurrence was also observed in association
with diabetes mellitus (HR, 1.23; P.04) and
hypertension (HR, 1.23; P.04). Of the noncardiovascular risk factors, only a history of

December 2014;89(12):1636-1643

http://dx.doi.org/10.1016/j.mayocp.2014.09.009
www.mayoclinicproceedings.org

RISK FACTORS FOR RECURRENT SPONTANEOUS EPISTAXIS

TABLE 3. Additional Potential Risk Factors for Recurrent Epistaxisa

Cases (No. [%]) Controls (No. [%])
(n461)
(n912)

Variable
Cardiovascular risk factors
Abdominal aortic aneurysm
Carotid artery stenosis
Chronic kidney disease
Congestive heart failure
Coronary artery disease
Cerebrovascular disease
Diabetes mellitus (types 1 and 2)
Hyperlipidemia
Hypertension
Hypothyroidism
Obesity
Peripheral vascular disease
Systemic lupus erythematosus
Noncardiovascular risk factors
Telangiectasias (skin and/or mucous membranes)
Hemangiomas
History of GI bleeding
Portal hypertension
History of anemia
Endometriosis
COPD
Migraines
Polyarteritis nodosa
Hyperthyroidism

HR (95% CI)

P valueb

14
33
29
130
154
68
125
239
310
74
123
63
3

(3.0)
(7.2)
(6.3)
(28.2)
(33.4)
(14.8)
(27.1)
(51.8)
(67.2)
(16.1)
(26.7)
(13.7)
(0.7)

37
86
65
176
310
145
214
515
608
168
250
109
9

(4.0)
(9.4)
(7.1)
(19.3)
(34.0)
(15.9)
(23.5)
(56.5)
(66.7)
(18.4)
(27.4)
(12.0)
(1.0)

1.17
0.86
1.18
1.61
1.20
1.09
1.23
0.99
1.23
1.08
1.05
1.30
0.57

(0.69-1.99)
.56
(0.60-1.22)
.39
(0.81-1.71)
.40
(1.32-1.98) <.001c
(0.99-1.45)
.07
(0.84-1.41)
.52
(1.01-1.52)
.04c
(0.82-1.19)
.91
(1.01-1.50)
.04c
(0.84-1.39)
.55
(0.85-1.29)
.67
(1.00-1.70)
.05
(0.18-1.77)
.33

7
15
91
1
146
11
92
35
3
15

(1.5)
(3.3)
(19.7)
(0.2)
(31.7)
(2.4)
(20.0)
(7.6)
(0.7)
(3.3)

20
48
173
4
278
21
188
81
4
49

(2.2)
(5.3)
(19.0)
(0.4)
(30.5)
(2.3)
(20.6)
(8.9)
(0.4)
(5.4)

0.95
0.65
1.15
0.63
1.34
1.21
1.12
1.13
2.04
0.93

(0.45-2.01)
(0.39-1.08)
(0.92-1.45)
(0.09-4.43)
(1.10-1.64)
(0.67-2.20)
(0.89-1.41)
(0.80-1.59)
(0.66-6.35)
(0.55-1.55)

.90
.10
.22
.64
.003c
.53
.32
.50
.22
.77

COPD chronic obstructive pulmonary disease; GI gastrointestinal; HR hazard ratio.

P values were derived from Cox proportional hazards regression modeling. The dependent variable is time to epistaxis recurrence. The
independent variables are each of the 13 dened cardiovascular risk factors and each of the 10 dened noncardiovascular risk factors.
c
Statistically signicant.
a

anemia was associated with a slight increase in

the risk of recurrence (HR, 1.34; P.003).
Several medications and some medication
combinations were also explored as risk factors for recurrent epistaxis (Table 4). Warfarin
use was associated with an increased risk of
recurrent epistaxis (HR, 1.43; P.001). In patients taking warfarin, concomitant aspirin use
was also associated with an increased risk of
epistaxis recurrence (HR, 1.46; P.01), but
this was not observed with aspirin use alone.
In contrast, clopidogrel use was not associated
with an increased risk of recurrent epistaxis
either alone or combined with aspirin and
warfarin. No signicant differences were
observed between cases and controls with
respect to nonsteroidal anti-inammatory drugs,
antidepressants, or vitamins. Of the herbal supplements assessed, ginseng was found be associated with an increased risk of epistaxis
recurrence (HR, 6.98; P.006), but use was
described infrequently.
Mayo Clin Proc. n December 2014;89(12):1636-1643
www.mayoclinicproceedings.org

Several laboratory values measured during

initial and subsequent epistaxis events or at
follow-up appointments were compared between cases and controls (Table 5). No signicant differences in median hemoglobin level,
platelet count, or INR were observed between
cases and controls. Similarly, no signicant differences were observed between the 2 groups
with respect to INR ranges. Subtherapeutic
INRs (<2.0) were found to be the most prevalent in patients taking warfarin.
Adverse outcomes within 30 days of the
initial epistaxis event were identied, but low
numbers of patients experiencing serious
adverse events precluded most statistical analyses (Table 6). The number of blood product
infusions was not found to be signicantly
different between cases and controls. Two cases
experienced new strokes with no history of
cerebrovascular disease. Of 3 controls who
experienced anginal symptoms, 2 had a history
of coronary artery disease and had temporarily

http://dx.doi.org/10.1016/j.mayocp.2014.09.009

1639

MAYO CLINIC PROCEEDINGS

TABLE 4. Medications and Herbal Supplementsa

Cases (No. [%])
(n461)

Variable

Antiplatelet and anticoagulant medications

Warfarin
127 (27.5)
Aspirin
190 (41.2)
Clopidogrel
25 (5.4)
Aspirin clopidogrel
19 (4.1)
Warfarin aspirin
51 (11.1)
Warfarin clopidogrel
4 (0.9)
Warfarin clopidogrel aspirin
2 (0.4)
Nonsteroidal anti-inammatory drugs
Ibuprofen
25 (5.4)
Naproxen
8 (1.7)
Celecoxib
4 (0.9)
Antidepressants
Citalopram
5 (1.1)
Escitalopram
3 (6.5)
Fluoxetine
9 (2.0)
Paroxetine
9 (2.0)
Sertraline
6 (1.3)
Herbal supplements and vitamins
Garlic
2 (0.4)
Ginseng
2 (0.4)
Omega-3 fatty acids
9 (2.0)
Saw palmetto
2 (0.5)
Vitamin C
28 (6.1)
Vitamin E
38 (8.2)

Controls (No. [%])

(n912)
179
403
51
41
78
7
6

(19.6)
(44.2)
(5.6)
(4.5)
(7.7)
(0.8)
(0.7)

HR (95% CI)
1.43
0.96
1.25
1.07
1.46
0.97
0.73

(1.17-1.75)
(0.80-1.16)
(0.85-1.83)
(0.68-1.70)
(1.09-1.96)
(0.36-2.58)
(0.18-2.92)

P valueb
.001c
.67
.26
.76
.01c
.94
.66

51 (5.6)
20 (2.2)
4 (4.4)

0.89 (0.59-1.33)
1.01 (0.50-2.02)
1.48 (0.55-3.97)

.56
.99
.43

21
3
20
13
26

(2.3)
(3.3)
(2.2)
(1.4)
(2.9)

0.98
1.72
0.73
1.19
0.62

(0.40-2.36)
(0.55-5.36)
(0.38-1.41)
(0.61-2.30)
(0.28-1.39)

.96
.35
.35
.61
.25

8
2
36
4
52
75

(0.9)
(0.2)
(4.0)
(0.4)
(5.7)
(8.2)

0.41
6.98
0.59
0.69
1.24
1.16

(0.10-1.62)
(1.73-28.15)
(0.31-1.15)
(0.17-2.79)
(0.85-1.83)
(0.83-1.62)

.20
.006c
.12
.61
.26
.38

HR hazard ratio.
P values were derived from Cox proportional hazards regression modeling. The dependent variable is time to epistaxis recurrence. The
independent variables are each of the 6 dened antiplatelet and anticoagulant medications and combinations, each of the 3 dened
nonsteroidal anti-inammatory drugs, each of the 5 dened antidepressants, and each of the 6 dened herbal supplements and vitamins.
c
Statistically signicant.
a

stopped taking their antiplatelet medications

despite these symptoms and 1 was newly diagnosed as having coronary artery disease.
DISCUSSION
Epistaxis is a common condition that is most
frequently considered a nuisance, and few episodes are severe enough to require medical
attention. However, a subset of patients with
epistaxis experience frequent recurrent episodes. Although several risk factors for single
epistaxis events have been identied, factors
related to epistaxis recurrence have not been
assessed. Herein, we examined this question
in a retrospective cohort study of patients
with either single or recurrent epistaxis events.
Basic demographic factors and traditional
risk factors for epistaxis were not related to
recurrence. A statistically signicant difference
in age was noted between the 2 groups, but
the mean age of cases was only 0.3 years
1640

Mayo Clin Proc.

(approximately 4 months) higher than that

of controls. None of the traditional risk factors
known to increase the risk of individual episodes of epistaxis were associated with risk
of recurrence. In theory, anatomical abnormalities might be thought to cause more frequent
episodes of epistaxis. The negative ndings in
this study suggest that this was not the case for
episodes severe enough to warrant medical
attention; however, their effect on minor nosebleeds that were not captured in this study
cannot be ascertained. Inammatory and infectious processes were also not found to be
risk factors, likely because no further epistaxis
would occur once they had resolved.
Several cardiovascular risk factors were
assessed to determine whether atherosclerosis
of the nasal vessels plays a role in recurrent
epistaxis. Of these, hypertension and diabetes
mellitus were found to be minor risk factors
for recurrence. Both of these diseases are known

December 2014;89(12):1636-1643

http://dx.doi.org/10.1016/j.mayocp.2014.09.009
www.mayoclinicproceedings.org

RISK FACTORS FOR RECURRENT SPONTANEOUS EPISTAXIS

to induce atherosclerotic changes in the vasculature systemwide. Other cardiovascular risk

factors, such as coronary artery disease, cerebrovascular disease, and peripheral vascular disease,
were not found to be associated with recurrent
epistaxis. A possible explanation for this observation is that hypertension and diabetes mellitus
are promoters of atherosclerosis, whereas coronary artery disease, cerebrovascular disease,
and peripheral vascular disease are, in fact, consequences of this process.
One of the more surprising ndings was the
high association of CHF with recurrent epistaxis.
Congestive heart failure is a risk factor for
epistaxis that is seldom mentioned in the literature and is likely underappreciated. The mechanism has been proposed to be related to
increased venous pressure in the nasal vessels.15
The results of this study suggest that patients
with a history of CHF may be more at risk for
recurrent nosebleeds of venous origin, in
contrast to arterial bleeding originating from
the Kiesselbach plexus on the nasal septum.
Another potential explanation for the association between CHF and recurrent epistaxis in
this case cohort may be that some of these patients have undiagnosed HHT. There is a known
association between high-output heart failure
and HHT due to right-to-left shunting of blood
through systemic arteriovenous malformations
(AVMs), particularly in the liver.16 These
AVMs cause reduced systemic vascular resistance, leading to neurohumeral activation and
resulting uid accumulation, presenting similarly to low-output CHF.17 Liver AVMs can be
diagnosed by Doppler ultrasound and treated
with embolization.16 Therefore, it may be prudent to consider evaluating for HHT in patients
with CHF who have been experiencing chronic
recurrent epistaxis over several years.
Of the noncardiovascular risk factors
examined, only a history of anemia was associated with recurrent epistaxis. However, the
median hemoglobin level at the time of presentation was within the reference range
for cases and controls, and no difference
was observed regarding the frequency of
blood transfusions in the 2 groups. This suggests that patients with recurrent epistaxis
may be more at risk for anemia at some point
in their lives, potentially even as a result of
chronic recurrent epistaxis. Other sources of
bleeding might also contribute to the anemia
Mayo Clin Proc. n December 2014;89(12):1636-1643
www.mayoclinicproceedings.org

TABLE 5. Laboratory Valuesa,b

Cases
(n461)
Variable
Hemoglobin (g/dL)
Platelets (x109/L)
INR
<2.0
2.0-3.0
>3.0

No.

Median

373
13.6
368
237
226
1.6
129 (57%)
59 (26%)
38 (17%)

Controls
(n912)
No.

Median

756
13.7
732
225
400
1.4
243 (61%)
110 (21%)
47 (12%)

HR (95% CI)

P value

0.95 (0.90-1.01)
0.999 (0.998-1.000)
1.03 (0.93-1.13)

.11c
.09c
.61c
.21d

HR hazard ratio; INR international normalized ratio.

SI conversion factor: To convert hemoglobin values to g/L, multiply by 10.
c
P value was derived from Cox proportional hazards regression modeling. The dependent variable
is time to epistaxis recurrence. The independent variables are hemoglobin level, platelet count,
and INR.
d
P value was derived from Fishers exact test.
a

in patients who are taking blood thinners,

although no signicant difference was
observed between the 2 groups with respect
to a history of gastrointestinal bleeding.
This difference may have been too small to
detect given the relatively small sample size.
Several medications that impair hemostasis were investigated in this study; however,
warfarin was one of the only medications
associated with an increased risk of recurrent
epistaxis. The increased bleeding risk associated with warfarin was independent of the
INR, and only a few nosebleeds occurred
when the INR was supratherapeutic (17% of
cases and 12% of controls). This nding suggests that the risk of recurrent epistaxis associated with warfarin may be additive to other
risk factors. An increased risk of recurrence
was also found with the combination of
TABLE 6. Adverse Outcomes Within 30 Days of the Initial Epistaxis Eventa

Outcome

Cases
(No. [%])
(n461)

Controls
(No. [%])
(n912)

Blood transfusion
Fresh frozen plasma infusion
Myocardial infarction
Angina
Stroke
Transient ischemic attack

9 (2.0)
2 (0.4)
0
0
2 (0.4)
0

23 (2.5)
6 (0.7)
0
3 (0.3)
0
0

HR (95% CI)

P valueb

1.39 (0.66-2.94)
0.80 (0.20-3.20)

.38
.75

HR hazard ratio.
P values were derived from Cox proportional hazards regression modeling. The dependent
variable is time to epistaxis recurrence. The independent variables are blood transfusion and fresh
frozen plasma infusion.

http://dx.doi.org/10.1016/j.mayocp.2014.09.009

1641

MAYO CLINIC PROCEEDINGS

aspirin and warfarin, but this was likely due

to the effect of warfarin because aspirin alone
was not associated with an increased risk.
Clopidogrel was also not found to be associated with recurrent epistaxis. Despite clopidogrel being a potent antiplatelet agent, no
increased risk was demonstrated with clopidogrel combined with aspirin (dual antiplatelet therapy) or with aspirin and warfarin
(triple therapy). Nevertheless, it is possible
that the effect of these combinations on recurrent epistaxis may have been too small to
detect given the relatively small sample size.
In addition, ginseng was found to be associated with an increased risk; however, low
counts and wide 95% CIs bring into question
the validity of this result.
Overall, very few patients experienced a
major adverse cardiovascular event within 30
days of an epistaxis event. Of those who did,
3 were newly diagnosed as having an atherosclerotic disease process and 2 temporarily
stopped taking their antiplatelet medications
despite having anginal symptoms. This low
event rate suggests that patients generally
continue to take their antiplatelet medications
during bleeding episodes or that they may
stop them for a period short enough to be of
no consequence. Conversely, these adverse
outcomes may be underestimated because
only major adverse cardiovascular events after
the rst epistaxis event were considered;
epistaxis events in which patients self-treated
were not taken into account. In addition, the
event rates may have been too small to be
detected with this small study size. Regardless,
physicians should promote continued antiplatelet medication adherence in patients
with known atherosclerotic disease to prevent
major adverse cardiovascular events.
Generalizability of the results of this study
to other population groups should be undertaken with caution because the study group
was made up mostly of a homogeneous white
population located in central Wisconsin. The
risk factors described may not affect patients
of other races or ethnicities in the same
manner. The same can be said about populations from vastly different climates because
temperature and humidity can also inuence
the development of epistaxis.
Additional study limitations include its retrospective design and reliance on documentation of
1642

Mayo Clin Proc.

epistaxis episodes in the medical record. It was

also a relatively small study, which raises the
possibility that some risk factors may not
have been detected owing to low power.
Also, not all nosebleeds that occurred were
captured; only episodes severe enough to warrant seeking medical attention with documented ndings on physical examination
were considered. In addition, the study was
designed to follow up patients longitudinally
only until their rst recurrence, which may
have been a relatively short period within a
3-year window. Epistaxis events during time
frames before or after the 3-year window
were not considered.
CONCLUSION
Traditional risk factors for single episodes of
epistaxis were not found to increase the risk
of recurrence. Signicant risk factors identied
for recurrent epistaxis included CHF, hypertension, and diabetes mellitus. Patients with
recurrent epistaxis may be at risk for anemia.
Warfarin use was found to increase the risk
of recurrence independently of INR. Patients
with recurrent epistaxis were unlikely to experience major adverse cardiovascular events
within 30 days. Further studies are required
to better understand antiplatelet medication
adherence in these patients.
ACKNOWLEDGMENTS
We would like to thank our research coordinators Theresa Pritzl, Madalyn Minervini,
and Deanna Cole for performing the manual
abstraction and quality assurance of the controls; our programmers Mike Olsen and Crystal Jacobson for performing the electronic
abstraction and building the database used
for statistical analysis; Rachel Stankowski for
editing and Marie Fleisner for formatting the
manuscript; and Deb Kempf for overseeing
and coordinating this study, without whom
this research project would not have been
possible.
Abbreviations and Acronyms: AVM = arteriovenous malformation; CHF = congestive heart failure; HHT = hereditary
hemorrhagic telangiectasia; HR = hazard ratio; INR = international normalized ratio
Grant Support: This project was funded through grant support provided to Dr Abrich by the Marsheld Clinic

December 2014;89(12):1636-1643

http://dx.doi.org/10.1016/j.mayocp.2014.09.009
www.mayoclinicproceedings.org

RISK FACTORS FOR RECURRENT SPONTANEOUS EPISTAXIS

Research Foundation from its disease-specic funds thanks

to contributions from generous donors.
Correspondence: Address to Victor Abrich, MD, Department of Cardiovascular Disease, Mayo Clinic Arizona,
13400 E Shea Blvd, Scottsdale, AZ 85259 (abrich.victor@
mayo.edu).

REFERENCES
1. Pallin DJ, Chng Y, McKay MP, Edmond JA, Pelletier AJ,
Camargo CA Jr. Epidemiology of epistaxis in US emergency
departments, 1992-2001. Ann Emerg Med. 2005;46(1):
77-81.
2. Bertrand B, Eloy P, Rombaux P, Lamarque C, Watelet JB,
Collet S. Guidelines to the management of epistaxis. B-ENT.
2005;(suppl 1):27-41:quiz 42-43.
3. Marx JA, Hochberger RS, Walls RM. Rosens Emergency Medicine:
Concepts and Clinical Practice. 5th ed. St Louis, MO: Mosby Inc; 2002.
4. Benninger MS, Marple BF. Minor recurrent epistaxis: prevalence
and a new method for management. Otolaryngol Head Neck
Surg. 2004;131(3):317-320.
5. Melia L, McGarry GW. Epistaxis: update on management. Curr
Opin Otolaryngol Head Neck Surg. 2010;19(1):30-35.
6. Manes RP. Evaluating and managing the patient with nosebleeds. Med Clin North Am. 2010;94(5):903-912.

Mayo Clin Proc. n December 2014;89(12):1636-1643

www.mayoclinicproceedings.org

7. McGarry GW, Gatehouse S, Vernham G. Idiopathic epistaxis, haemostasis, and alcohol. Clin Otolaryngol Allied Sci. 1995;20(2):174-177.
8. Sparacino LL. Epistaxis management: whats new and whats
noteworthy. Lippincotts Prim Care Pract. 2000;4(5):498-507.
9. Melia L, McGarry G. Epistaxis in adults: a clinical review. Br J
Hosp Med (Lond). 2008;69(7):404-407.
10. Middleton PM. Epistaxis. Emerg Med Australas. 2004;16(5-6):
428-440.
11. Ibrashi F, Sabri N, Eldawi M, Belal A. Effect of atherosclerosis and hypertension on arterial epistaxis. J Laryngol Otol. 1978;92(10):877-881.
12. Pollice PA, Yoder MG. Epistaxis: a retrospective review of hospitalized patients. Otolaryngol Head Neck Surg. 1997;117(1):49-53.
13. Shehab N, Sperling LS, Kegler SR, Budnitz DS. National estimates of emergency department visits for hemorrhagerelated adverse events from clopidogrel plus aspirin and from
warfarin. Arch Intern Med. 2010;170(21):1926-1933.
14. Cox DR. Regression models and life-tables. J R Stat Soc Series B
Stat Methodol. 1972;34(2):187-220.
15. Kanowitz SJ, Citardi MJ, Batra PS. Contemporary management
strategies for epistaxis. In: Stucker FS, De Souza C, Lian TS,
Draf W, Schick B, eds. Rhinology and Facial Plastic Surgery. Leipzig, Germany: Springer; 2009:142.
16. Cottin V, Dupuis-Girod S, Lesca G, Cordier JF. Pulmonary
vascular manifestations of hereditary hemorrhagic telangiectasia
(rendu-osler disease). Respiration. 2007;74(4):361-378.
17. Mehta PA, Dubrey SW. High output heart failure. QJM. 2009;
102(4):235-241.

http://dx.doi.org/10.1016/j.mayocp.2014.09.009

1643

Copyright of Mayo Clinic Proceedings is the property of Frontline Medical Communications

and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.