Original Paper

Cerebrovasc Dis 2016;41:273–282 Received: August 16, 2015

Accepted: January 13, 2016
DOI: 10.1159/000444128
Published online: February 6, 2016

Increased Risk of Stroke in Patients with Isolated

Third, Fourth, or Sixth Cranial Nerve Palsies:
A Nationwide Cohort Study
Chu-Peng Hoi a Yung-Tai Chen b, c Jong-Ling Fuh d, e Chun-Pai Yang f, g
Shuu-Jiun Wang d, e
a
Department of Neurology, Institute of Internal Medicine, Centro Hospitalar Conde de São Januário, Macao,
b
National Yang-Ming University School of Medicine, c Department of Nephrology, Institute of Internal Medicine,
Taipei City Hospital Heping Fuyou Branch, d Department of Neurology, Neurological Institute, Taipei Veterans
General Hospital, and e Institute of Brain Science, National Yang-Ming University, Taipei, f Department of Neurology,
Kuang Tien General Hospital, and g Department of Nutrition, Huang-Kuang University, Taichung, Taiwan

Key Words
Oculomotor nerve palsy · Trochlear nerve palsy · Abducens
nerve palsy · Stroke · Risk factor · Cohort study
Abstract
Background and Purpose: The aim of this nationwide cohort
study was to evaluate whether the occurrence of isolated
3rd, 4th or 6th cranial nerve (CN) palsies is associated with a
higher risk of ischemic stroke. Methods: This study utilized
data from Taiwan Longitudinal Health Insurance Database
during 1995–2012. Subjects aged 20 years or older who had
isolated CN 3/4/6 palsies diagnosed by a neurologist or oph-
thalmologist between January 2000 and December 2011
were included. A set of propensity score matched, randomly
sampled patients who had never been diagnosed with CN
3/4/6 palsies were extracted to constitute the control group
(cases and controls = 1: 4). All subjects were followed until
death, loss due to follow-up or completion of the study. Cox
proportional hazard regression model stratified by matched
pairs was used to estimate the hazards ratio (HR) of ischemic
stroke. Results: A total of 657 patients with isolated CN 3/4/6
palsies (61.1% male, mean age 54.8 years) were identified.
Compared with control group, the patients with isolated CN
3/4/6 palsies exhibited an increased risk of ischemic stroke
(CN3: adjusted HR 3.69 (95% CI 2.20–6.19); CN4: 2.71 (95% CI
1.11–6.64); CN6: 2.15 (95% CI 1.31–3.52)). The association be-
tween CN 3/4/6 palsies and ischemic stroke was detected in
both separate subgroup and sensitivity analyses. Conclu-
sions: The patients with CN 3/4/6 palsies exhibited an in-
creased risk of developing ischemic stroke. Therefore, isolat-
ed ocular motor nerves palsies appear to represent an unrec-
ognized risk factor for ischemic stroke, and these require
further confirmation and exploration.
© 2016 S. Karger AG, Basel
Introduction
Except for paresis of the facial nerve, ocular motor cra-
nial nerves (CNs) are most frequently involved in mono-
neuropathies of the CNs, which involve CN3 (oculomo-
tor), CN4 (trochlear) and/or CN6 (abducens). The preva-
lence of CN 3/4/6 palsies has been documented in large
population-based studies, and cases involving isolated
ocular cranial nerve palsy (OCNP) have often been ob-
served [1, 2].
Drs. C.-P. Hoi and Y.-T. Chen contribute equally to this paper.

© 2016 S. Karger AG, Basel Dr. Shuu-Jiun Wang, MD

1015–9770/16/0416–0273$39.50/0 Department of Neurology, Neurological Institute
Taipei Veterans General Hospital
E-Mail karger@karger.com
Taipei 11217 (Taiwan)
www.karger.com/ced
E-Mail sjwang @ vghtpe.gov.tw
 Isolated OCNP has a variety of causes [3–5]. The most
widely postulated mechanism describes microvascular
ischemia to the CN due to atherosclerotic risk factors. It
has been reported that vascular causes account for 20% of
CN3 palsies, for 18.6% of CN4 palsies and for 17.7% of
CN6 palsies [6]. Other etiologies such as aneurysm, trau-
ma, neoplasm, brain stem infarction, infection, inflam-
mation, sinus thrombosis and multiple sclerosis have also
been proposed [7–10].
In addition to the above microvascular ischemia etiol-
ogy, emerging data suggest that some peripheral CN in-
juries or disorders are associated with a higher risk of
stroke, particularly those involving CN5, CN7 or CN8
[11–13]. However, to our knowledge, there have been no
reports regarding a possible association between risk of
ischemic stroke and isolated CN 3/4/6 palsies in adults
[14, 15]. Therefore, a nationwide cohort study was con-
ducted to assess the risk of ischemic stroke in patients
diagnosed with CN 3/4/6 palsies.
Subjects and Methods
Data Source
Records from the Longitudinal Health Insurance Database
(LHID) between 1995 and 2012 were obtained from the National
Health Insurance Research Database (NHIRD) by Taiwan’s Na-
tional Health Insurance Program. This program was launched in
1995, and it covers 99% of the 23 million people currently living in
Taiwan. The LHID consists of 1 million beneficiaries that were
randomly sampled from the original NHI beneficiaries. The data-
base includes the entire registry and claims data that stem from this
health insurance system, with the entries ranging from demo-
graphic data to detailed orders from ambulatory and inpatient
care. Several published papers have used the NHIRD as the basis
for their studies. Moreover, diseases in the database are coded ac-
cording to the International Classification of Disease, Ninth Revi-
sion, Clinical Modification (ICD-9-CM) diagnosis codes. This
study was initiated following approval by the Institutional Review
Board of Taipei Veterans General Hospital in Taiwan (IRB 2015-
03-002CC).
Study Design
The goal of this population-based, observational study was to
assess the potential association between patients diagnosed with
isolated CN 3/4/6 palsies and their risk of ischemic stroke. Two
cohorts were enrolled, including a cohort of patients diagnosed
with CN 3/4/6 palsies and a control cohort. The definitions of CN
3/4/6 palsies were ICD-9 codes 378.51, 378.52 for CN3, 378.53 for
CN4 and 378.54 for CN6. The CN 3/4/6 palsies cohort consisted
of patients that had an ambulatory visit or hospitalization coded
with the above ICD-9-CM coding between January 2000 and
December 2011. The date of diagnosis of the CN 3/4/6 palsies was
defined as the index date. According to previous studies, the eti-
ologies of patients with multiple CN 3/4/6 palsies were most often
274 Cerebrovasc Dis 2016;41:273–282
DOI: 10.1159/000444128
due to inflammations such as Tolosa–Hunt Syndrome or tumor
[4]. In the current study, a history of cerebrovascular disease was
defined as patients with any hospitalization or outpatient visits for
cerebrovascular disease (ICD-9 codes 430.x-438.x) before the in-
dex date. Each subject in the cohorts had at least 5 years of look-
back period. And according to the civil law in Taiwan, the defini-
tion of an adult is a person who is ≥20 years of age. Therefore,
Institutional Review Boards in Taiwan define patients of <20 years
of age as pediatric study subjects, who are vulnerable and require
a higher level of ethical review. Thus, patients who were <20 years
of age, had a history of cerebrovascular disease and had multiple
CN 3/4/6 palsies on the index date were excluded from this study.
The control cohort was extracted from the remaining patients in
the LHID, and it consisted of patients who had never been diag-
nosed with CN 3/4/6 palsies. Considering the latter, the index date
for the control cohort was randomly selected from the index date
of 1 patient in the CN 3/4/6 palsies cohort. In addition, the same
exclusion criteria that were applied to the 3/4/6 CN palsies cohort
were applied to the control cohort. The 3 most common primary
discharge diagnoses for the control cohorts were ischemic stroke
(<1%), diabetes mellitus (<1%) and pneumonia (<1%). The con-
trol subjects were not a group of any special diseases. They were
randomly selected from the general population according to the
propensity score for diagnosis of CN palsies. The validation of the
diagnoses of CN 3/4/6 palsies was just published elsewhere [16]. In
short, the diagnostic codes for CN3, CN4 and CN6 palsies were
confirmed in 209 of 212 patients, 126 of 129 patients and 261 of
266 patients after review of the medical records, yielding a positive
predictive value of 98.6, 97.7 and 98.1%, respectively.
Propensity Score Matching
For each subject, the propensity scores for the likelihood of di-
agnosis of CN 3/4/6 palsies were calculated from baseline covari-
ates using multivariate logistic regression analysis. Baseline demo-
graphics included age, gender, year of index date, month of index
date, monthly income, urbanization levels, outpatient visits in the
past year and Charlson comorbidity index (CCI) value [17]. Data
related to the use of anti-hypertension (HTN) drugs as a metric of
blood pressure control were also collected. The concomitant use
of other cardiovascular-associated medications was also deter-
mined, including antiplatelet agents, warfarin, nitrates, statins, di-
pyridamole, steroids, estrogen/progesterone, non-steroidal anti-
inflammatory drugs, proton-pump inhibitors and anti-hypergly-
cemic drugs. Furthermore, other systemic diseases and risk factors
of ischemic stroke that were not included in the CCI were exam-
ined, such as HTN, asthma, valvular heart disease, dyslipidemia,
autoimmune disease and drug abuse. For each patient in the CN
3/4/6 palsies cohort, 4 control patients with the most similar un-
derlying conditions were identified and matched according to the
closest propensity score based on nearest-neighbor matching
without replacement and by using calipers of width equal to 0.1 SD
of the logit of the propensity score.
Outcome
The primary outcome for this study was the occurrence of isch-
emic stroke during the follow-up period. The occurrence of isch-
emic stroke was defined as hospitalization with a principal diagno-
sis of ischemic stroke (ICD-9-CM code 433.x, 434.x or 436). Both
cohorts were followed until death, until loss due to follow-up or
until completion of the study (December 31, 2012).
Hoi/Chen/Fuh/Yang/Wang
 Statistical Analyses
Descriptive statistics were used to describe the baseline char-
acteristics of the 2 cohorts. Baseline characteristics of the matched
groups were compared using standardized differences. The pro-
pensity scores for the likelihood of diagnosis of CN 3/4/6 palsies
were calculated using a multivariate logistic regression analysis,
conditional on the baseline covariates. The p value of c-statistic
for this model was 0.8. The incidence rates of the outcome be-
tween the 2 groups were calculated using Poisson distributions.
The cumulative incidence of events was calculated using the Ka-
plan–Meier method and was compared between the 2 groups us-
ing the log-rank test. Cox proportional hazard regression model
was used to estimate the hazards ratio (HR) of ischemic stroke. In
order to avoid the bias, the model was stratified on matched pairs.
The likelihood ratio test was used to test the interactions between
CN 3/4/6 palsies and the following conditions: gender, age, HTN,
diabetes mellitus, hyperlipidemia and coronary artery disease.
Subgroup analyses were also performed accordingly. Finally, sen-
sitivity analyses were performed with different inclusion criteria
to validate the association between CN 3/4/6 palsies and risk of
ischemic stroke.
A Microsoft SQL Server 2012 (Microsoft Corp., Redmond,
Wash., USA) was used for data linkage, processing and sampling.
Propensity scores were calculated with SAS version 9.3 (SAS Cam-
pus Drive, Cary, N.C., USA). All other statistical analyses were
conducted using STATA statistical software (version 13.0; Stata-
Corp, College Station, Tex., USA). Statistical significance was de-
fined as a p value <0.05.
Results
Characteristics of the 2 Cohorts Examined
For each patient with CN 3/4/6 palsies (n = 657), there
were 4 matched control subjects (n = 2,628). In the former
group of patients with isolated OCNP, 225 patients had
CN3 palsy (34.2%), 118 had CN4 palsy (18.0%) and 314
had CN6 palsy (47.8%). The mean age of the CN 3/4/6
palsies cohort was 54.8 years (SD 15.8 years). A majority
of the isolated OCNP patients were men (61.0%) and had
CCI scores <3 (62.1%). The most common comorbidity
for the CN 3/4/6 palsies group was HTN (45.9%), fol-
lowed by peptic ulcer disease (37.3%), chronic pulmonary
disease (33.0%), dyslipidemia (32.3%) and diabetes
(32.0%). Both the CN 3/4/6 palsies cohort and the control
cohort had similar baseline characteristics (tables 1 and
2).
Risk of Ischemic Stroke
The mean follow-up period was 5.8 years and a total of
151 patients developed ischemic stroke. Of these patients,
57 were in the CN 3/4/6 palsies cohort (among 3,391 per-
son-years) and 94 were in the control cohort (among
15,276 person-years; fig. 1–4), and the corresponding in-
Increased Risk of Stroke in Patients with
Isolated 3rd, 4th, or 6th CN Palsies
cidence rates of ischemic stroke were 16.81 and 6.15 per
1,000 person-years, respectively. Thus, the patients with
CN 3/4/6 palsies had a significantly higher risk of isch-
emic stroke (adjusted HR 2.74, 95% CI 1.97–3.81; table 3),
with the adjusted HR values compared with the matched
subjects being 3.69 (95% CI 2.20–6.19) for the CN3 palsy
patients, 2.71 (95% CI 1.11–6.64) for the CN4 palsy pa-
tients and 2.15 (95% CI 1.31–3.52) for the CN6 palsy pa-
tients. We used the likelihood ratio tests to test the inter-
actions for different CNPs and ischemic stroke. The re-
sults showed that there was no interaction between
different CNPs and ischemic stroke (pinteraction = 0.319).
Subgroup and Sensitivity Analyses of the Risk of
Ischemic Stroke
Tests of interaction failed to reach statistical signifi-
cance in all of the subgroup analyses that were performed
to investigate a possible association between CN 3/4/6
palsies and ischemic stroke according to gender, age,
HTN, diabetes mellitus, hyperlipidemia and coronary ar-
tery disease (table 4). In the sensitivity analyses that were
performed with different inclusion criteria, the Cox mod-
el revealed similar results (table 5). In fact, we used the
Cox model rather than logistic regression to analyze the
risk of ischemic stroke between groups. Thus, we tested
the proportional hazards assumption on the basis of
Schoenfeld residuals. The p value was 0.248.
Discussion
For the present cohort, during a mean follow-up pe-
riod of 5.8 years, isolated ocular motor nerve palsy was
associated with a 2.74-fold increased risk of ischemic
stroke. This association was confirmed in the various sub-
group and sensitivity analyses that were performed. These
data suggest that isolated OCNP is a previously unrecog-
nized risk factor for ischemic stroke, and they strengthen
the hypothesis that certain CN palsies are associated with
a higher incidence of ischemic stroke [11–13].
A major strength of the present study was the use of a
nationwide population-based database. This database en-
abled us to identify incident cases of stroke in patients
with OCNP and to establish a temporal relationship. In
addition, the large sample size provided adequate statisti-
cal power for evaluating a relatively infrequent disorder,
such as isolated OCNP. Third, an age- and propensity
score-matched control cohort was used to minimize se-
lection bias and the influence of medical care-seeking be-
havior.
Cerebrovasc Dis 2016;41:273–282 275
DOI: 10.1159/000444128
 Table 1. Demographics and clinical characteristics of the patients examined

Characteristics Propensity score-matched

OCNP cohort control cohort standardized
difference†

Number of patients 657 2,628

Mean age, years (SD) 54.8 (15.8) 55.1 (16.7) –0.017
Male, n (%) 401 (61.0) 1,623 (61.8) –0.011
Monthly income, n (%)
Dependent 175 (26.6) 725 (27.6) –0.024
NT 0–19,100 137 (20.9) 523 (19.9) 0.021
NT 19,100–42,000 286 (43.5) 1,168 (44.4) –0.016
>NT 42,000 59 (9.0) 212 (8.1) 0.037
Urbanization, n (%)*
Level 1 240 (36.5) 932 (35.5) 0.019
Level 2 382 (58.1) 1,574 (59.9) –0.032
Level 3 30 (4.6) 101 (3.8) 0.035
Level 4 5 (0.8) 21 (0.8) –0.005
Outpatient visits within the previous 12 months of diagnosis, n (%)
0−10 190 (28.9) 729 (27.7) 0.023
11−20 195 (29.7) 782 (29.8) –0.005
21−30 126 (19.2) 494 (18.8) 0.007
31−40 71 (10.8) 300 (11.4) –0.021
>40 78 (11.9) 323 (12.3) –0.015
Outpatient neurology visits within the previous 12 months of diagnosis, n (%)
0 531 (80.8) 2,112 (80.4) 0.002
1−5 144 (21.9) 451 (17.2) 0.003
6−10 10 (1.5) 48 (1.8) –0.024
>10 5 (0.8) 17 (0.6) 0.013
CCI, n (%)
0 172 (26.2) 774 (29.5) –0.076
1 139 (21.2) 523 (19.9) 0.029
2 100 (15.2) 398 (15.1) 0.000
≥3 249 (37.9) 933 (35.5) 0.046
Concomitant medications, n (%)
Anti-HTN agents
ACE inhibitor or ARB 75 (11.4) 346 (13.2) –0.005
Alpha-blockers 9 (1.4) 34 (1.3) 0.006
Beta-blockers 82 (12.5) 347 (13.2) –0.023
Calcium-channel blocker 109 (16.6) 413 (15.7) 0.026
Diuretics 39 (5.9) 120 (4.6) 0.060
Other anti-HTN drug 11 (1.7) 38 (1.4) 0.018
Anti-hyperglycemic drug 119 (18.1) 433 (16.5) 0.053
Antiplatelet agents‡ 57 (8.7) 202 (7.7) 0.035
Dipyridamole 24 (3.7) 98 (3.7) –0.005
Estrogen or progesterone 12 (1.8) 55 (2.1) –0.020
Nitrate 15 (2.3) 58 (2.2) 0.004
NSAID 157 (23.9) 656 (25.0) –0.024
Proton-pump inhibitor 4 (0.6) 22 (0.8) 0.008
Statin 28 (4.3) 101 (3.8) 0.020
Steroid 62 (9.4) 251 (9.6) –0.000
Warfarin 2 (0.3) 5 (0.2) 0.023
Coexisting conditions, n (%)
AIDS 1 (0.2) 1 (0.0) 0.037
Asthma 63 (9.6) 250 (9.5) 0.001
Atrial fibrillation 10 (1.5) 44 (1.7)
Autoimmune disease 22 (3.3) 108 (4.1) –0.033
Cancer 83 (12.6) 329 (12.5) 0.015

276 Cerebrovasc Dis 2016;41:273–282 Hoi/Chen/Fuh/Yang/Wang

DOI: 10.1159/000444128
 Table 1. (continued)

Characteristics Propensity score-matched

OCNP cohort control cohort standardized
difference†

Chronic kidney disease 74 (11.3) 281 (10.7) 0.017

Chronic liver disease 172 (26.2) 662 (25.2) 0.030
Chronic pulmonary disease 217 (33.0) 871 (33.1) 0.001
Coronary artery disease 163 (24.8) 659 (25.1) 0.002
Dementia 8 (1.2) 33 (1.3) –0.004
Diabetes 210 (32.0) 795 (30.3) 0.044
Drug abuse 16 (2.4) 65 (2.5) –0.003
Dyslipidemia 212 (32.3) 820 (31.2) 0.026
Heart failure 43 (6.5) 153 (5.8) 0.029
HTN 308 (46.9) 1,239 (47.1) –0.004
Myocardial infarction 16 (2.4) 58 (2.2) 0.014
Peptic ulcer disease 245 (37.3) 966 (36.8) 0.017
Peripheral vascular disease 26 (4.0) 97 (3.7) 0.013
Plegia 8 (1.2) 24 (0.9) 0.055
Rheumatoid disease 23 (3.5) 92 (3.5) –0.001
Valvular heart disease 45 (6.8) 195 (7.4) –0.023
† Imbalance defined as an absolute value >0.108.
‡ Including aspirin, clopidogrel, ticlopidine and cilostazol.
* Urbanization levels in Taiwan are divided into 4 strata according to the Taiwan National Health Research
Institute in which level 1 is referred to as the ‘most urbanized’ and level 4 as the ‘least urbanized’.
NT  = New Taiwan dollar; PPI = proton-pump inhibitor; ACE = angiotensin-converting enzyme; ARB =
angiotensin-II receptor blocker;  NSAID = non-steroidal anti-inflammatory drug.

Table 2. Demographics and clinical characteristics of the OCNP patients examined

Characteristics CN3 palsy CN4 palsy CN6 palsy p value

Number of patients 225 118 314

Mean age, years (SD) 55.3 (16.6) 54.7 (15.1) 54.4 (15.5) 0.787
Male, n (%) 136 (60.4) 81 (68.6) 184 (58.6) 0.158
Number of patients that experienced a stroke during
the follow-up period, n (%) 26 (11.6) 8 (6.8) 23 (7.3) 0.164
Mean age at stroke, years (SD) 66.3 (9.8) 61.5 (11.4) 64.3 (14.4) 0.591
Mean duration between development of ischemic stroke
and diagnosis of OCNP, years, mean (95% CI) 3.2 (1.9–4.6) 3.2 (0.6–5.8) 3.9 (2.4–5.4) 0.776

The finding that CN6 palsy (47.8%) was the most com-
monly isolated OCNP, followed by CN3 palsy (34.2%)
and CN4 palsy (18.0%), is consistent with the results of
previously published large-scale series [2, 18], and it sup-
ports the validity of our study. In addition, the validation
study of the diagnosis of OCNP in 2 hospitals demon-
strated that a high accuracy rate was achieved. In combi-
nation with a previous validation study of patients with
ischemic stroke [19], the NHIRD appears to be a valid
resource for population research studies of the associa-
tion between OCNP and ischemic stroke.
There are several possible pathophysiological process-
es that may explain the higher risk of ischemic stroke that
was observed after a diagnosis of isolated OCNP. First, it
is likely that the common arteriosclerotic causes of both
conditions not only result in microvascular ischemia, but
also contribute to the occurrence of subsequent ischemic
stroke [20, 21]. A retrospective comparative study previ-

Increased Risk of Stroke in Patients with Cerebrovasc Dis 2016;41:273–282 277

Isolated 3rd, 4th, or 6th CN Palsies DOI: 10.1159/000444128
 0.24 OCNP
Control
log-rank test
p < 0.001

Cumulative incidence of stroke

0.16

0.08

0
Fig. 1. Cumulative incidence of ischemic 0 2 4 6 8 10 12
stroke among OCNP patients and matched Follow-up period (years)
control subjects. Cumulative risk of devel- No. of patients at risk
oping ischemic stroke over time for the OCNP cohort 657 512 351 243 156 86 20
OCNP patients group (solid line) and con- Matched cohort 2,628 2,197 1,607 1,127 795 441 120
trol group (dash line).

0.24 CN3 palsy

Control
log-rank test
p < 0.001
Cumulative incidence of stroke

0.16

0.08

0
Fig. 2. Cumulative incidence of ischemic 0 2 4 6 8 10 12
stroke among CN3 palsy patients and Follow-up period (years)
matched control subjects. Cumulative risk No. of patients at risk
of developing ischemic stroke over time for OCNP cohort 225 171 125 83 48 27 8
the CN3 palsy patients group (solid line) Matched cohort 900 740 551 377 258 150 50
and control group (dash line).

ously reported that arteriosclerosis was the main cause of
lacunar brain infarcts and ischemic ocular motor nerve
palsies [22]. In addition, traditional cardiovascular risk
factors (e.g., HTN, diabetes and hyperlipidemia) have
been found to commonly affect patients with isolated
OCNP [23]. A postmortem examination of an ischemic
infarct within the center of the ocular motor nerve trunk
showed an occluded nutrient artery [24], and this is con-
sistent with the hypothesis that generalized arterioscle-
rotic change in the intracranial arteries, including the

278 Cerebrovasc Dis 2016;41:273–282 Hoi/Chen/Fuh/Yang/Wang

DOI: 10.1159/000444128
 0.24 CN4 palsy
Control
log-rank test
p = 0.023

Cumulative incidence of stroke

0.16

0.08

0
Fig. 3. Cumulative incidence of ischemic 0 2 4 6 8 10 12
stroke among CN4 palsy patients and Follow-up period (years)
matched control subjects. Cumulative risk No. of patients at risk
of developing ischemic stroke over time for OCNP cohort 118 96 67 47 37 16 5
the CN4 palsy patients group (solid line) Matched cohort 472 392 285 201 141 86 20
and control group (dash line).

0.24 CN6 palsy

Control
log-rank test
p = 0.002
Cumulative incidence of stroke

0.16

0.08

0
Fig. 4. Cumulative incidence of ischemic 0 2 4 6 8 10 12
stroke among CN6 palsy patients and Follow-up period (years)
matched control subjects. Cumulative risk No. of patients at risk
of developing ischemic stroke over time for OCNP cohort 314 245 159 113 71 43 7
the CN6 palsy patients group (solid line) Matched cohort 1,256 1,065 771 549 396 205 50
and control group (dash line).

supplied nutrient arteries of CN 3/4/6, may contribute to
the occurrence of OCNP and ischemic stroke.
It is also important to note that there may be other
mechanisms involved that are independent of the com-
mon cardiovascular risk factors or age-related vascular
changes that have been associated with ischemic stroke.
In the subgroup analyses performed in the present study,
isolated OCNP was identified as an independent risk fac-
tor for stroke even after adjusting for patient age and the
presence of common cardiovascular risk factors such as

Increased Risk of Stroke in Patients with Cerebrovasc Dis 2016;41:273–282 279

Isolated 3rd, 4th, or 6th CN Palsies DOI: 10.1159/000444128
 Table 3. Incidence rates and risk of ischemic stroke among CN 3/4/6 palsies cohort and the matched control
cohort

OCNP cohort Control cohort Propensity score matched

number person- incidence number person- incidence HR** (95% CI) p value
of events years rate* of events years rate*

CN 3/4/6 palsies 57 3,391 16.81 94 15,276 6.15 2.74 (1.97–3.81) <0.001

CN3 palsy 26 1,139 22.83 32 5,190 6.17 3.69 (2.20–6.19) <0.001
CN4 palsy 8 663 12.06 12 2,735 4.39 2.71 (1.11–6.64) 0.029
CN6 palsy 23 1,589 14.48 50 7,351 6.80 2.15 (1.31–3.52) 0.002

* Per 103 person-years; ** adjusted for propensity score.

Table 4. Subgroup analyses of risk of stroke among the OCNP cohort and the control cohort

Characteristics HR* (95% CI) p value Interaction

p value

Gender
Male 2.56 (1.66–3.93) <0.001 0.508
Female 3.15 (1.87–5.29) <0.001
Age, years
20–55 4.29 (1.90–9.69) <0.001 0.192
≥56 2.52 (1.75–3.63) <0.001
HTN
Yes 2.67 (1.84–3.85) <0.001 0.630
No 3.29 (1.56–6.91) 0.002
Diabetes mellitus
Yes 2.66 (1.68–4.24) <0.001 0.855
No 2.98 (1.85–4.78) <0.001
Hyperlipidemia
Yes 2.95 (1.76–4.96) <0.001 0.833
No 2.75 (1.79–4.22) <0.001
Coronary artery disease
Yes 2.80 (1.72–4.54) <0.001 0.992
No 2.89 (1.83–4.56) <0.001

* Adjusted for propensity score.

Table 5. Sensitivity analyses of association between risk of ischemic stroke and OCNP

HR* (95% CI) p value

Excluding with follow-up period <90 days 2.24 (1.56–3.21) <0.001

Excluding with follow-up period <180 days 2.26 (1.57–3.26) <0.001
Excluding with follow-up period <365 days 2.37 (1.62–3.46) <0.001
Excluding patients with history of aneurysm, SAH, Graves’ disease, head injury,
Tolosa–Hunt syndrome, pseudotumor cerebri or nasopharyngeal cancer 3.20 (2.00–5.12) <0.001

* Adjusted for propensity score.

280 Cerebrovasc Dis 2016;41:273–282 Hoi/Chen/Fuh/Yang/Wang

DOI: 10.1159/000444128
HTN, diabetes, etc. Other possible mechanisms have
been proposed, and these include roles for inflammation,
infectious processes and hypercoagulability [6, 8, 25–27].
These factors were not evaluated in the present study, but
would be factors to consider in future studies.
Two clinical implications of the present findings are as
follows. First, with isolated OCNP identified as a risk fac-
tor for ischemic stroke, a more intensive preventative fo-
cus may be necessary after the occurrence of an isolated
OCNP. Second, the significant association between iso-
lated OCNP and ischemic stroke suggests that a common
mechanism underlies both diseases. It remains to be de-
termined whether the underlying mechanism of this as-
sociation is due to microvascular ischemia.
There were several limitations associated with the
present study. First, the diagnoses of isolated ocular mo-
tor palsy, stroke and other comorbidities were entirely
dependent on the ICD codes from the NHIRD database
and, therefore, on the diagnostic accuracy of the data-
base. The bureau of Taiwan NHI has made every effort
to verify the accuracy of the diagnoses in the database
[28] although the sensitivity and specificity of the valida-
tion study may differ in different levels of hospitals [19].
However, the diagnoses of ischemic stroke and OCNP,
the 2 important disease entities were validated previous-
ly. Second, the original NHIRD database does not in-
clude information pertaining to smoking habits, alcohol
intake, body mass index, physical activity and metabolic
profiles, and these may affect risk estimations of ischemic
stroke. Third, the severity and type of stroke experienced
cannot be precisely extracted from ICD codes, and this
prevented further subgroup analysis. Fourth, the mean
duration of the follow-up period was only 5.8 years, and
this did not allow us to evaluate the long-term effects of
isolated OCNP on stroke. Fifth, in our database, the dis-
ease was coded according to the ICD-9-CM. Thus, it is
impossible to confirm the pupil involvement in those
with CN 3 palsy; the latter was associated with diabetes
mellitus and was more likely to be expected in the isch-
emic patients.
Summary/Conclusions
In conclusion, the present nationwide population-
based cohort study demonstrates a significantly increased
risk of ischemic stroke following isolated OCNP. There-
fore, this previously unrecognized risk factor for ischemic
stroke warrants further confirmation and investigation.
Sources of Funding
This study was supported in part by grants from Taipei Veter-
ans General Hospital (VGHUST104-G7-1-1, V104C-082,
V104E9-001), Ministry of Science and Technology of Taiwan
(MOST 104-2314-B-010-015-MY2, and MOST 103-2321-B-010-
017), Ministry of Science and Technology support for the Center
for Dynamical Biomarkers and Translational Medicine, National
Central University, Taiwan (MOST 103-2911-I-008-001), Aca-
demia Sinica (Grant No. IBMS-CRC103-P04), Brain Research
Center, National Yang-Ming University, Ministry of Health and
Welfare, Taiwan (MOHW 103-TDU-B-211-113-003, MOHW
104-TDU-B-211-113-003, MOHW 105-TDU-B-211-113-003),
and a grant from the Ministry of Education, Aim for the Top Uni-
versity Plan.
Disclosure Statement
S.-J. Wang has served on the advisory boards of Allergan and
Eli Lilly Taiwan. He has received speaking honoraria from the
Taiwan branches of Pfizer, Eli Lilly and GlaxoSmithKline. He has
received research grants from the Taiwan National Science Coun-
cil, Taipei Veterans General Hospital and the Taiwan Headache
Society. J.-L. Fuh is a member of the scientific advisory board of
Novartis and has received research support from the Taiwan Na-
tional Science Council and Taipei Veterans General Hospital.

References
1 Patel SV, Mutyala S, Leske DA, Hodge DO,
Holmes JM: Incidence, associations, and eval-
uation of sixth nerve palsy using a popula-
tion-based method. Ophthalmology 2004;
111:369–375.
2 Richards BW, Jones FR Jr, Younge BR: Causes
and prognosis in 4,278 cases of paralysis of the
oculomotor, trochlear, and abducens cranial
nerves. Am J Ophthalmol 1992;113:489–496.
3 Tamhankar MA, Biousse V, Ying GS, Prasad
S, Subramanian PS, Lee MS, Eggenberger E,
Moss HE, Pineles S, Bennett J, Osborne B,
Volpe NJ, Liu GT, Bruce BB, Newman NJ,
Galetta SL, Balcer LJ: Isolated third, fourth,
and sixth cranial nerve palsies from presumed
microvascular versus other causes: a prospec-
tive study. Ophthalmology 2013; 120: 2264–
2269.
4 Berlit P: Isolated and combined pareses of
cranial nerves III, IV and VI. A retrospective
study of 412 patients. J Neurol Sci 1991; 103:
10–15.
5 Bendszus M, Beck A, Koltzenburg M, Vince
GH, Brechtelsbauer D, Littan T, Urbach
H,  Solymosi L: MRI in isolated sixth
nerve palsies. Neuroradiology 2001; 43: 742–
745.
6 Rucker CW: The causes of paralysis of
the  third, fourth and sixth cranial nerves.
Am  J Ophthalmol 1966; 61(5 pt 2):1293–
1298.
7 Munakata A, Ohkuma H, Nakano T, Shi-
mamura N: Abducens nerve pareses associ-
ated with aneurysmal subarachnoid hemor-
rhage. Incidence and clinical features. Cere-
brovasc Dis 2007;24:516–519.

Increased Risk of Stroke in Patients with Cerebrovasc Dis 2016;41:273–282 281

Isolated 3rd, 4th, or 6th CN Palsies DOI: 10.1159/000444128
 8 Thömke F, Gutmann L, Stoeter P, Hopf HC:
Cerebrovascular brainstem diseases with iso-
lated cranial nerve palsies. Cerebrovasc Dis
2002;13:147–155.
9 Rush JA, Younge BR: Paralysis of cranial
nerves III, IV, and VI. Cause and prognosis in
1,000 cases. Arch Ophthalmol 1981;99:76–79.
10 Lavin PJ, Donahue SP, Jacobson DM, Moster
ML, Galetta SL, Liu GT, Eggenberger ER: Iso-
lated trochlear nerve palsy in patients with
multiple sclerosis. Neurology 2000; 55: 321–
322.
11 Pan SL, Chen LS, Yen MF, Chiu YH, Chen
HH: Increased risk of stroke after trigeminal
neuralgia – a population-based follow-up
study. Cephalalgia 2011;31:937–942.
12 Lee CC, Su YC, Chien SH, Ho HC, Hung SK,
Lee MS, Chou P, Chiu BC, Huang YS: In-
creased stroke risk in Bell’s palsy patients
without steroid treatment. Eur J Neurol 2013;
20:616–622.
13 Lin HC, Chao PZ, Lee HC: Sudden sensori-
neural hearing loss increases the risk of stroke:
a 5-year follow-up study. Stroke 2008; 39:
2744–2748.
14 Jung JS, Kim DH: Risk factors and prognosis
of isolated ischemic third, fourth, or sixth cra-
nial nerve palsies in the Korean population. J
Neuroophthalmol 2015;35:37–40.
15 Rowe F, VIS Group UK: Prevalence of ocular
motor cranial nerve palsy and associations
following stroke. Eye (Lond) 2011; 25: 881–
887.
282 Cerebrovasc Dis 2016;41:273–282
DOI: 10.1159/000444128
16 Yang CP, Chen YT, Fuh JL, Wang SJ: Mi-
graine and risk of ocular motor cranial nerve
palsies: a nationwide cohort study. Ophthal-
mology 2015;123:191–197.
17 Charlson ME, Pompei P, Ales KL, MacKenzie
CR: A new method of classifying prognostic
comorbidity in longitudinal studies: develop-
ment and validation. J Chronic Dis 1987; 40:
373–383.
18 Akagi T, Miyamoto K, Kashii S, et al: Cause
and prognosis of neurologically isolated third,
fourth, or sixth cranial nerve dysfunction in
cases of oculomotor palsy. Jpn J Ophthalmol
2008;52:32–35.
19 Cheng CL, Kao YH, Lin SJ, et al: Validation of
the national health insurance research data-
base with ischemic stroke cases in Taiwan.
Pharmacoepidemiol Drug Saf 2011; 20: 236–
242.
20 López-Cancio E, Matheus MG, Romano JG,
Liebeskind DS, Prabhakaran S, Turan TN,
Cotsonis GA, Lynn MJ, Rumboldt Z, Chi-
mowitz MI: Infarct patterns, collaterals and
likely causative mechanisms of stroke in
symptomatic intracranial atherosclerosis.
Cerebrovasc Dis 2014;37:417–422.
21 Tadokoro Y, Sakaguchi M, Yamagami H,
Okazaki S, Furukado S, Matsumoto M, Miwa
K, Yagita Y, Mochizuki H, Kitagawa K: Echo-
genicity of medium-to-large carotid plaques
predicts future vascular events. Cerebrovasc
Dis 2014;38:354–361.
22 Pollak L, Kessler A, Rabey MJ, et al: Clinical
characteristics of patients with ischemic ocu-
lar nerve palsies and lacunar brain infarcts: a
retrospective comparative study. Acta Neurol
Scand 2005;111:333–337.
23 Reddy PS, Reddy RC, Satapathy M: Aetiolog-
ical study of the third, fourth and sixth cra-
nial nerve paralysis. Indian J Ophthalmol
1972;20:159–163.
24 Dreyfus PM, Hakim S, Adams RD: Dia-
betic ophthalmoplegia; report of case, with
postmortem study and comments on vas-
cular supply of human oculomotor nerve.
AMA Arch Neurol Psychiatry 1957; 77: 337–
349.
25 Sheu JJ, Chiou HY, Kang JH, Chen YH, Lin
HC: Tuberculosis and the risk of ischemic
stroke: a 3-year follow-up study. Stroke 2010;
41:244–249.
26 Kang JH, Ho JD, Chen YH, et al: Increased
risk of stroke after a herpes zoster attack: a
population-based follow-up study. Stroke
2009;40:3443–3448.
27 Mwanza JC, Ngweme GB, Kayembe DL: Ocu-
lar motor nerve palsy: a clinical and etiologi-
cal study. Indian J Ophthalmol 2006;54:173–
175.
28 Wen CP, Tsai SP, Chung WS: A 10-year ex-
perience with universal health insurance in
Taiwan: measuring changes in health and
health disparity. Ann Intern Med 2008; 148:
258–267.
Hoi/Chen/Fuh/Yang/Wang