PAIN 155 (2014) 26992706

www.elsevier.com/locate/pain

Early evoked pain or dysesthesia is a predictor of central poststroke pain

Henriette Klit a,b,, Anne P. Hansen a, Ninna S. Marcussen a, Nanna B. Finnerup a, Troels S. Jensen a,b
a
b

Danish Pain Research Center, Aarhus University, Aarhus, Denmark

Department of Neurology, Aarhus University Hospital, Aarhus, Denmark

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

a r t i c l e

i n f o

Article history:
Received 29 June 2014
Received in revised form 26 September
2014
Accepted 29 September 2014

Keywords:
Central neuropathic pain
Central poststroke pain
Pain
Predictors
Stroke

a b s t r a c t
Central poststroke pain (CPSP) is a central neuropathic pain condition caused by a cerebrovascular lesion
affecting the central somatosensory nervous system. Once developed, CPSP is difcult to treat, so there is
an interest in identifying stroke patients at risk for the development of CPSP. This study examined if sensory abnormalities, including evoked dysesthesia, allodynia, or hyperalgesia to static and dynamic touch,
cold, and pinprick, at stroke onset are a predictor for the development of CPSP. Consecutive stroke
patients were recruited from a large prospective study of poststroke pain in Aarhus, Denmark, between
2007 and 2008. Patients underwent a structured pain interview and a standardized sensory examination
within 4 days of admission, and a structured telephone interview was conducted after 3 and 6 months.
Patients who developed poststroke pain in the affected side without any other plausible cause were classied as having possible CPSP. A total of 275 stroke patients completed the study, and 29 patients (10.5%)
were classied as having possible CPSP. The diagnosis was conrmed by a clinical examination in 15 of 17
patients, corresponding to a prevalence of 8.3%. The presence of allodynia, hyperalgesia, or dysesthesia in
response to the sensory examination at stroke onset increased the odds for CPSP at 6 months by 4.6 (odds
ratio; 95% condence interval 1.513.9). The combination of reduced or absent sensation to pinprick or
cold and early evoked pain or dysesthesia at onset increased odds by 8.0 (odds ratio; 95% condence
interval 2.624.8). In conclusion, early evoked pain or dysesthesia is a predictor for CPSP.
2014 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction
Central poststroke pain (CPSP) is a central neuropathic pain
condition caused by a vascular lesion of the central somatosensory
nervous system [16]. It is characterized by pain and sensory abnormalities in the body parts that correspond to the injured brain area.
The incidence of this devastating condition, which is associated
with both reduced quality of life and increased mortality [7,27],
is estimated to be 7% to 8% in stroke survivors [1,19]. The condition
is often refractory to treatment, and it is thus important to identify
patients who are at risk for the development of CPSP in order to follow these patients more carefully, inform them about central pain,
initiate treatment, and conduct prophylactic treatment studies in
the future.
The underlying mechanisms of CPSP are not known, but abnormal spinothalamocortical tract function [23,34], loss of inhibition
Corresponding author at: Danish Pain Research Center, Aarhus University
Hospital, Norrebrogade 44, Building 1A, DK-8000 Aarhus C, Denmark. Tel.: +45 7846
3287; fax: +45 7846 3269.
E-mail address: henriette.klit@clin.au.dk (H. Klit).

[6,9,11,12,14], central sensitization [37], and neuroplastic changes

[24,28,29] are all mechanisms that have been implicated in this
and other central neuropathic pain conditions. Central sensitization is dened as an increased responsiveness of nociceptive neurons in the central nervous system (CNS) to their normal or
subthreshold afferent input [1].
Neuropathic pain is characterized by negative and/or positive
symptoms and signs. It may be spontaneous (ongoing or intermittent) or evoked, and it may be associated with nonpainful dysesthesia or paresthesia [5,34]. The negative signs in CPSP include
loss or reduction of thermal and pinprick sensibility, while positive
symptoms and signs are pain or unpleasantness evoked by touch,
pressure, or thermal stimuli. This dual combination of loss and
hypersensitivity in the affected area is usually explained by loss
of input to a population of CNS neurons participating in the processing of somatosensory information, and at the same time hyperexcitability in the same or other neurons within the CNS [14,19].
Our previous studies have shown that sensory hypersensitivity
in the affected body parts may occur already at onset of stroke
[1,19], but it is unclear if this sensory hypersensitivity presenting
as hyperalgesia, dysesthesia, or allodynia is a risk factor for the

http://dx.doi.org/10.1016/j.pain.2014.09.037
0304-3959/ 2014 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

2700

H. Klit et al. / PAIN 155 (2014) 26992706

subsequent development of pain. In 2 recent studies, early sensory

hypersensitivity was found to predict later onset of central neuropathic pain (below-level pain) in patients with spinal cord injury
[10,38]. In this prospective study, we aimed to determine if stroke
patients with early evoked pain or dysesthesia, dened as ndings
of allodynia, hyperalgesia, or dysesthesia in response to the sensory examination at stroke onset, have an increased risk of developing CPSP.
2. Methods
2.1. Patient recruitment
We included stroke patients admitted consecutively to the
Stroke Unit of the Department of Neurology, Aarhus University
Hospital, Aarhus, Denmark, between February and September
2007 and between February and July 2008 (Fig. 1). Inclusion criteria
were age of 18 years or above, informed consent, and a diagnosis of
stroke according to the World Health Organization criteria (ICD-10
codes: I61, I63, I64.9, I67.6, and I67.7). Exclusion criteria were a
diagnosis of trasient ischemic attack (TIA) (G45.9) or subarachnoid

hemorrhage (I60.9), inability to communicate, severe dementia, or

pronounced somnolence. Results on pain development are reported
elsewhere [13].
All included patients underwent a structured interview and a
standardized bedside sensory examination within the rst 4 days
of admission (initial examination), including sensory testing for dysesthesia, allodynia, and hyperalgesia. Dysesthesia is dened as an
unpleasant abnormal sensation, whether spontaneous or evoked;
allodynia as pain due to a stimulus that does not normally provoke
pain; and hyperalgesia as increased pain from a stimulus that normally provokes pain [26]. To determine the presence of these phenomena, we used static and dynamic touch, cold, and pinprick
stimulation (in the following called evoked pain or dysesthesia)
[13]. The interview included questions about pain conditions before
the stroke and the presence of spontaneous and evoked pain or dysesthesia at or after stroke onset. Medical records from the hospital
admission and results of computed tomography (CT) and/or magnetic resonance imaging (MRI) scans were obtained. All initial examinations were done by 1 of 2 investigators (APH or NSM).
A structured follow-up telephone interview was conducted 3
and 6 months after stroke onset. The interview included questions

Fig. 1. Flow chart of study.

2701

H. Klit et al. / PAIN 155 (2014) 26992706

about the status of the pain reported in the previous interview,

pain experienced within the last week, and the presence of spontaneous and evoked pain or dysesthesia. Only patients who completed the 6-month follow-up were included in the study.
After the 6-month follow-up, patients were classied as having
possible CPSP, other poststroke pain, or no poststroke pain on the
basis of the data from the initial examination and the interviews
at the admission and the 3- and 6-month follow-up. Information
about early evoked dysesthesia or pain was not used to classify
the patients. Patients were identied as having possible CPSP
on the basis of the proposed grading system for neuropathic pain
[20,33] if they fullled the following 3 criteria: (1) onset of pain
at or after stroke onset, (2) pain located in the stroke-affected side
of the body, and (3) no other plausible cause of the pain (including
poststroke shoulder pain isolated to the shoulder region) [13]. The
affected side was conrmed post hoc on the basis of information
from medical records on symptoms and signs at stroke onset and
CT or MRI results. If the affected side could not be identied with
certainty (n = 10), both sides were included as affected. Early
evoked pain or dysesthesia was considered relevant if: (1) it was
unilateral or asymmetrical, (2) it was found in the affected side
of the body, and (3) its onset was after the stroke, or if onset was
uncertain.
In order to conrm the diagnosis, all patients identied with
possible CPSP were invited to participate in a clinical examination
(follow-up examination). The examination included a standardized
interview, a bedside sensory examination, and thermal sensory
testing. Patients were interviewed about medical history, medication, and pain, which was assessed with 4 items from the Brief Pain
Inventory (general activity, mood, sleep, and enjoyment of life) [8]
and a Danish version of the 10-item DN4 [4], and patients were
asked to complete a pain drawing. On the basis of all the available
information, except for information on early evoked pain or dysesthesia, patients were classied as having denite CPSP or no CPSP
according to the proposed grading system for neuropathic pain
[33]. All follow-up examinations were performed by a physician
(HK) who had not participated in the initial examination. Patients
who were classied as having denite CPSP based on the grading
system for neuropathic pain at the follow-up examination, including patients who fullled the criteria for denite neuropathic pain
but in whom other causes of pain could not be excluded, are
referred to as denite CPSP patients in the following.
2.2. Sensory examination
At both examinations, a cotton ball was used to examine for static touch sensation, a von Frey monolament (monolament No.
5.88; Semmes Weinstein Touch Test, Stoelting, IL, USA) for pinprick
sensation, a brush (Somedic, Hrby, Sweden) for dynamic touch,
and a metal roller for cold sensation (initial examination: a metal
roll at 20 C [Somedic, Hrby, Sweden]; follow-up examination:
metal roll at room temperature).
Sensation to each stimulus was noted as absent, diminished,
normal, or increased compared to the unaffected mirror body part.
In those cases where the affected side could not be determined on
the basis of the sensory examination alone, the history and the
result of the brain scan determined the affected side. Sensory ndings were indicated on a body chart. The presence of evoked dysesthesia, allodynia, and hyperalgesia to each stimulus was noted, and
the intensities of evoked pain and dysesthesia were scored on an
11-point (0 = no pain, 10 = worst possible pain) numeric rating
scale (NRS). The initial examination was conducted bilaterally on
the front of the forearm, thenar, cheek, shin, and the dorsum of
the foot, while the entire front of the body, or in a few cases (for
practical reasons) only parts of the body, were examined at the follow-up examination.

At the clinical follow-up examination, sensory thermal testing

(TSA 2001; Medoc Thermotest, Ramat Yishai, Israel) was conducted
with a 3  3 cm2 thermode, a baseline temperature of 32 C, a rate
of change of 1 C/s for detection and 5 C/s for pain thresholds, and
cutoff values at 0 C and 50 C. Cold detection threshold (CDT),
warm detection threshold (WDT), cold pain threshold (CPT), and
heat pain threshold (HPT) were determined on the affected side
(in an area of pain, either cheek, forearm, or shin) and on the corresponding contralateral side (average of 3 measurements). A side
difference of P2 C in detection thresholds, P10.3 C in cold pain
thresholds, and P4.2 C in heat pain thresholds were considered
abnormal [30].
2.3. Ethics
The study was conducted in accordance with the Helsinki Declaration. Written and oral information was given, and informed
consent was obtained before both the initial and the follow-up
examination. The study was approved by the regional research ethics committee (20060116) and by the Danish Data Protection
Agency (2006-41-6900).
2.4. Statistical analysis
Statistical analysis was performed by Intercooled Stata software, version 9.2 (StataCorp, College Station, TX, USA). Data are
presented as mean and standard deviation (SD) with 95% condence intervals (CI) or as median with range where appropriate.
Parametric data were analyzed by Students t test. Nonparametric
data were analyzed using by Mann-Whitney and Kruskal-Wallis
(rank sum) tests. Dichotomous data were analyzed by Pearsons
chi-square (v2) test and Fishers exact test. Odds ratios (OR) are
presented with 5% and 95% CI. P values less than .05 were considered statistically signicant.
3. Results
In total, 299 stroke patients were included in the prospective
study, and 275 patients (153 men, 122 women) completed the 6month follow-up (Fig. 1). The mean age of the patients was 65.6
(SD 13.0) years. Detailed information including stroke characteristics are reported elsewhere [13]. After the 6-month follow-up
interview, 29 patients (11 men) were classied as having possible
CPSP, corresponding to an incidence of 10.5% (mean age 60.4 [SD
9.39] years) [13]. In 25 of these 29 patients, the pain was located
in an area with sensory abnormalities conrmed by the initial
examination at stroke onset. Possible CPSP was more common
after lesions affecting the left side compared with the right (20/
138 vs 8/126, P = .032).
Seventeen of 29 patients with possible CPSP agreed to participate in the follow-up examination. In the remaining 12 patients,

Table 1
Number of patients (N = 275) with evoked pain or dysesthesia and median (range)
intensity on the NRS (010).
Stimulus

Brush
Light touch
Cold thermo rolla
Pinprick
Total (n = 42)

Evoked pain

Evoked dysesthesia

NRS

12
5
21
9
33

4
4
5
4

0
0
3
12
13

NRS

5 (49)
5 (110)

(18)
(25)
(110)
(25)

a
Only 1 patient with cold-evoked pain and 4 with cold-evoked dysesthesia also
had brush-evoked dysesthesia, suggesting that cold-evoked hypersensitivity is not
only related to the touch of the thermal roll.

2702

H. Klit et al. / PAIN 155 (2014) 26992706

Table 2
OR for the development of CPSP.a
Item

Predictor

95%
95%
CI

Sens
(%)

4.6

1.5
13.9

10 (4.1%)

3.6

0.7
18.3
0.7
18.3
1.7
31.6
0.6
59.3
0.4
8.3
1.0
15.1
0.6
8.8
0.3
21.2
3.7
46.6
5.9
633
2.1
18.6
1.5
12.3
1.3
15.9
0.8
6.2
2.1
17.7
1.8
104.6
2.0
26.1
3.7
46.6
2.6
24.8
NA

Denite CPSP
(n = 15), n (%)

Not suspected of
CPSP (n = 246), n (%)

Primary outcome
1
Early evoked pain or dysesthesia at examination

6 (40%)

31 (12.6%)

Secondary outcomes
1a
Early evoked pain (allodynia or hyperalgesia) at examination

2 (13%)

OR

1b

Early evoked pain or dysesthesia to 1 or more stimuli

2 (13%)

10 (4.1%)

3.6

Early pain or dysesthesia to brush at examination

3 (20%)

8 (3.3%)

7.4

Early pain or dysesthesia to touch at examination

1 (7%)

3 (1.2%)

5.7

Early pain or dysesthesia to cold at examination

2 (13%)

20 (8.1%)

1.7

Early dysesthesia or hyperalgesia to pinprick at examination

3 (20%)

15 (6.1%)

3.9

Reported spontaneous pain or unpleasantness at onset

3 (20%)

24 (9.8%)

2.3

2a

Reported spontaneous pain* at onset

1 (7%)

7 (2.9%)

2.4

Reported pain or dysesthesia evoked by touch or cold at onset

5 (33%)

9 (3.7%)

13.2

3a

Reported pain evoked by touch or cold at onset

3 (20%)

1 (0.4%)

61.3

2 or 3

Reported spontaneous pain or unpleasantness or reported

dysesthesia or pain evoked by touch or cold at onset
Reported spontaneous pain or unpleasantness or reported or
evoked dysesthesia or pain at onset
Reported or evoked pain

7 (47%)

30 (12.2%)

6.3

8 (53%)

52 (21.1%)

4.3

4 (27%)

18 (7.3%)

4.6

7 (47%)

71 (28.9%)

2.2

Early hyperesthesia at examination (including evoked

dysesthesia or pain)
Early anesthesia at examination

8 (53%)

39 (15.9%)

6.1

Hypo- or anesthesia to 1 or more stimuli at examination

14 (93%)

125 (50.8%)

13.6

6a

Hypo- or anesthesia to cold or pinprick at examination

12 (80%)

88 (35.8%)

7.2

1 and 5

Early evoked pain/dysesthesia and anesthesia at examination

5 (33%)

9 (3.7%)

13.2

1 and 6a

Early evoked pain/dysesthesia and reduced/abolished sensation

to cold or pinprick at examination
Abnormal sensory ndings at examination, either hyper-, hypo, or anesthesia

6 (40%)

19 (7.7%)

8.0

15 (100%)

149 (60.6%)

1, 2 or 3
1a or 2a
4

4 or 6

NA

Spec
(%)

PPV
(%)

Acc
(%)

16

96

40

85

17

95

13

91

17

95

13

91

27

95

20

92

25

95

93

95

13

87

17

95

20

90

11

95

20

86

13

95

92

36

96

33

93

75

95

20

95

19

96

47

85

13

97

53

77

18

95

27

89

96

47

70

17

97

53

82

10

99

93

52

12

98

80

64

36

96

33

93

24

96

40

89

100

100

43

OR = odds ratio; CPSP = central poststroke pain; CI = condence interval; PPV = positive predictive value; sens = sensitivity; spec = specicity; acc = accuracy.
*
Reported spontaneous pain not including pain due to headache or evoked pains.
a
Comparison between reported symptoms and sensory ndings on the initial examination at stroke onset correlated to the later classication as denite CPSP, conrmed
by a clinical examination (n = 15) compared to patients not suspected of CPSP (n = 246).

pain had disappeared in 2 patients, 8 patients declined participation, and 2 patients were lost to follow-up (Fig. 1). The median
time from the stroke to the follow-up examination was
32.4 months (range 6.236.6 months). At the follow-up examination, the patients were classied as follows: denite (n = 15,
including 4 patients in whom other causes of pain could not
be excluded), not CPSP (n = 1, poststroke shoulder pain), and
no longer pain (CPSP-like dysesthesia, n = 1). In total, 15 of the
17 examined patients were classied as having denite CPSP.
This corresponds to a minimum prevalence of clinically conrmed denite CPSP in 5.9% of patients within 6 months after
stroke. If patients in whom other causes of pain could not be
excluded are omitted, the corresponding prevalence is 4.3%. If
we expect those without response to have the same prevalence,
it corresponds to a prevalence of 8.3% (15/19  29/275). The sensory ndings and reported symptoms and pain of the 15 patients
with denite CPSP are summarized in Fig. 3 and Supplementary
Tables 13.

3.1. Predictors of CPSP

Denite CPSP conrmed by the clinical follow-up examination
was considered to be the outcome measure. The 15 patients with
denite CPSP were compared with the 246 (275 29) patients
without possible CPSP (Fig. 1).
3.2. Primary outcome measure: early evoked dysesthesia or pain
Early evoked pain or dysesthesia to brush, touch, cold, or pinprick stimulation was found in 53 of 275 patients. Relevant evoked
pain or dysesthesia was identied in 42 patients (15.3%), while the
remaining 11 ndings were nonrelevant, ie, bilateral and symmetrical (n = 3), in the unaffected side (n = 6), or had onset before
stroke (n = 2). Of these 42 patients, 13 had allodynia or hyperalgesia, and 29 patients had only evoked dysesthesia. The ndings and
the intensity of evoked pain or dysesthesia to the 4 stimuli are
summarized in Table 1.

H. Klit et al. / PAIN 155 (2014) 26992706

Of the 42 patients with early evoked pain or dysesthesia, 5 were

later classied as having possible CPSP but were not available for
follow-up, 31 were not suspected of having CPSP, and 6 were later
classied as having denite CPSP (Fig. 1). Early evoked pain or dysesthesia was found in 37.9% (11 of 29) of the patients with possible
CPSP compared to 12.6% (31 of 246) of the patients not suspected
of having CPSP (P < .001, v2). This corresponds to an OR of 4.2 (95%
CI 1.89.8) of developing possible CPSP in patients with early
evoked pain or dysesthesia. In patients with denite CPSP, early
evoked pain or dysesthesia was seen in 6 (40.0%) of 15, corresponding to an OR of 4.6 (95% CI 1.513.9), with a positive predictive
value of 40%, a sensitivity of 16%, a specicity of 96%, and an accuracy of 85% (Table 2, item 1; Fig. 2). Of the 6 patients with early
evoked pain or dysesthesia who later developed denite CPSP, 1
patient had pinprick dysesthesia (NRS 5), 1 had pinprick hyperalgesia (NRS not done), 1 had cold allodynia (NRS 4), 1 had brush
dysesthesia (NRS 7), 1 had a combination of dysesthesia to brush
(NRS 4), touch (NRS 4), and pinprick (NRS 4), and 1 had a combination of dysesthesia to brush (NRS 8) and cold (NRS 3) (Supplementary Table 1).
The presence of early evoked pain or dysesthesia in combination with reduced or absent sensation to cold or pinprick increased
the odds for central poststroke pain at 6 months by 8.0 (OR; 95% CI
2.624.8) (Table 2, items 1 and 6a).
3.3. Secondary outcome measures: other symptoms and signs
Secondary outcomes are presented in Table 2. Patients who
reported pain evoked by touch or cold at onset were more likely
to be classied as having denite CPSP at follow-up (OR 61.3;
95% CI 5.9633) (Table 2, item 3a). Patients with decreased or
absent sensation, ie, hypo- or anesthesia to 1 or more stimuli, were
also more likely to be classied as having CPSP (OR 13.6; 95% CI
1.8104.6) (Table 2, item 6). There was also a signicant correlation between the presence of early anesthesia and CPSP (OR 6.1;
95% CI 2.117.7) (Table 2, item 5). Patients with reduced or absent
sensation to pinprick or cold, indicating abnormal spinothalamocortical tract function, were also more likely to be classied with
CPSP (OR 7.2; 95% CI 2.026.1) (Table 2, item 6a; Fig. 3).
We found no correlation between the presence of early hyperesthesia and the development of CPSP (OR 2.2; 95% CI 0.86.2)
(Table 2, item 4). Reported spontaneous pain or unpleasantness
at the initial interview (item 2) was not signicantly correlated
to CPSP, whereas there was a signicant correlation between

Fig. 2. Frequency of early evoked dysesthesia or pain in patients with and without
CPSP at 6-month follow-up. P < .001.

2703

reported touch- or cold-evoked dysesthesia or pain in the affected

side and CPSP (OR 13.2; 95% CI 3.746.6) (Table 2, item 3). The
combination of early evoked pain and dysesthesia and absent sensation to 1 or more stimuli at the time of the initial examination
increased the OR for developing CPSP to 13.2 (95% CI 3.746.6)
with a sensitivity of 36% and a specicity of 96% (accuracy 93%)
(Table 2, items 1 and 5).
4. Discussion
Our previous studies have indicated that CPSP is more common
in patients with somatosensory decits [1,19]. In this prospective
study, we found that the presence of evoked dysesthesia, allodynia,
or hyperalgesia in the affected side of the body within 4 days of
stroke onset was a predictor for the development of CPSP within
6 months after stroke onset (OR 4.6; 95% CI 1.513.9). Other early
predictors include patient-reported evoked pain and dysesthesia
and sensory abnormalities at the initial examination.
A hallmark of neuropathic pain is the combination of negative
and positive signs [15]. Importantly, we found that the combination of early evoked pain or dysesthesia and absent sensation to
1 or more stimuli at the time of the initial examination increased
the OR for developing CPSP (OR 13.2; 95% CI 3.746.6).
It has previously been shown that evoked dysesthesia and allodynia are present in most patients with CPSP [2,3,5], but to our
knowledge, ours is the rst study to compare early sensory ndings to the subsequent development of CPSP in a prospective manner. We found that early evoked pain or dysesthesia was present
within 4 days of stroke onset in 15% of the patients. Findings of
evoked dysesthesia or pain may in some, but not all, cases reect
neuronal hyperexcitability, ie, an excessive reaction of the neurons
to stimuli [18]. In some patients, these symptoms were present
early or immediately after stroke onset. The presence of dysesthesia and allodynia in CPSP patients may reect central sensitization
[11], which has been proposed as one of the mechanisms underlying this pain condition [17,20,34]. Consistent with the present ndings, recent studies have shown that early sensory hypersensitivity
and allodynia are predictors for below-level central pain in
patients with spinal cord injury. Zeilig et al. [38] found that
dynamic mechanical allodynia, temporal summation of pain, and
hyperpathia predicted central pain, while Finnerup et al. [10]
found that brush, cold, warm, and single or repetitive pinprick predicted central pain, of which cold-evoked pain or dysesthesia was
the best predictor. The authors suggested that this neuronal hyperexcitability, which may develop after damage to the spinothalamic
tracts, develops gradually and precedes the development of spontaneous central pain [10,38]. In line with these ndings, we found
that early pain or dysesthesia to brush and the combination of
early evoked pain or dysesthesia and decreased or absent sensation
to cold or pinprick, which may be interpreted as an indication of
damage to the spinothalamic tracts, signicantly predicted development of central pain.
The presence of early evoked pain or dysesthesia is not a perfect
predictor for the development of CPSP. The sensitivity, specicity,
and accuracy of early evoked dysesthesia or pain was 16%, 96%,
and 85%, respectively, indicating that other mechanisms are
involved. In accordance with the present study, not all patients
in the studies on spinal cord injury pain who developed central
pain had early sensory hypersensitivity.
A high prevalence of CPSP has been reported after lesions affecting the posterior parts of the thalamus and the lateral medulla
[22,25,32]. While the study by Sprenger et al. [32] focused on
structural changes within the thalamus as a risk factor for CPSP,
this study found that early clinical hypersensitivity in the body
part corresponding to the affected brain lesion represents a
predictor for subsequent development of CPSP. It is possible that

2704

H. Klit et al. / PAIN 155 (2014) 26992706

Fig. 3. Area of sensory abnormalities at onset (left) compared to areas of sensory abnormalities (middle) and pain (right) at follow-up examination in patients with denite
CPSP.

a combination of early sensory ndings combined with information on the localization of the lesion, such as specic areas within
the thalamus [32], may prove to be an even better predictor. Laserevoked potentials or functional neuroimaging could also be of
value in identifying high-risk patients in future preemptive treatment studies. To date, only one study has been published on prophylactic treatment of CPSP [22]. In this prophylactic study, a

total of 39 patients with thalamic stroke were randomized to

receive either amitriptyline or placebo. Only 7 of 39 patients developed CPSP; unfortunately, the study did not have sufcient statistical power to show any benecial effect.
In the present study, we found an estimated minimum prevalence of clinically conrmed denite CPSP in 5.9% of patients
within 6 months after stroke and a prevalence of denite CPSP

H. Klit et al. / PAIN 155 (2014) 26992706

without other possible causes of pain in 4.3%. It can be difcult to

determine if a patient with poststroke pain has CPSP. We applied
clearly dened criteria in order to do this. The prevalence found
in this study corresponds to other population-based studies using
this and other methods of identication [1,19,35]. Unlike some of
the previous epidemiologic studies on CPSP, patients with dysesthesia only were not included in the present study. If these patients
had been included, the reported prevalence of CPSP would have
been higher.
The present study has some shortcomings. Twelve of 29
patients with possible CPSP were not available for the clinical follow-up examination. However, because CPSP was conrmed by the
clinical examination in 15 of the 17 patients with possible CPSP, we
assume that the frequency of CPSP was the same in the patients
who did not participate in the follow-up examination. This corresponds to an estimated minimum prevalence of denite CPSP in
8.3% (15/19  27/275), and in 5.7% (11/19  27/275) of denite
CPSP patients in whom other causes of pain could be excluded,
in line with previously published estimates of the prevalence of
CPSP [1,19].
It may be argued that patients with early evoked dysesthesia or
pain have CPSP with early onset, and that the signs simply reect a
disease that is already present. However, only 1 patient had spontaneous pain at onset (Supplementary Table 1).
In order not to tire out the acute stroke patients, the initial sensory examination was only performed in the face and in the distal
limbs. Thus, we do not have detailed topographical data. Information on the topographical distribution of sensory ndings could be
an additional source of predictors.
It was not always possible retrospectively to determine
whether a sensory nding was caused by the present stroke, a previous stroke, or other neurologic disease such as neuropathy. In
cases of uncertainty, the sensory ndings were categorized as
new. CPSP due to a previous stroke was not evaluated in this study
because only pain with onset after the current stroke was
described. Although all patients classied as denite CPSP fullled
the diagnostic criteria according to the proposed grading system
for neuropathic pain, we could not clinically rule out that other
(peripheral) diseases were contributing to the pain in a small subset of patients. These patients were classied as patients with CPSP
in whom other causes of pain could not be excluded. Including
these patients in the study implies a risk of overestimating the
prevalence of CPSP.
The last follow-up in the prospective study was at 6 months.
The onset of CPSP is within 6 months after stroke in the majority
of patients [1,36]. In the study by Andersen [1], only 3 of 207
patients developed CPSP after 6 months. However, there are case
reports in the literature indicating a later onset, and it is therefore
possible that we have overlooked cases with late onset. We cannot
rule out the notion that some of the patients with poststroke
shoulder pain may have had a localized type of CPSP in the shoulder region. Thus, there is a risk of underreporting the incidence of
CPSP [21,31]. Interestingly, at the time of follow-up, pain had disappeared in 3 of the 29 patients with possible CPSP. This conrms
our previous nding that CPSP may diminish or disappear over
time in some patients [18].
Relevant early evoked pain or dysesthesia was found in 15% of
the stroke patients. Early evoked pain or dysesthesia was found
to be a signicant factor for the development of CPSP, indirectly
suggesting that central sensitization may be an underlying pathophysiologic mechanism. Because of the increased risk, a close follow-up of patients with early evoked pain or dysesthesia is
recommended in order to identify and treat potential pain. Future
clinical studies on the prevention of this pain condition could be
based on patients with early hypersensitivity, preferably if other
factors related to CPSP can be identied.

2705

Conict of interest
The authors report no conict of interest.
Acknowledgments
The authors thank Helle O. Andersen for editing help. This work
was supported by the Velux Foundation, the Danish Medical Association Research Fund, a Hoejmosegaard Grant, and a scholarship
from Aarhus University. The funding sources had no involvement
or role in the collection, analysis, or interpretation of the data or in
the writing of this article or the decision to submit it for publication.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.pain.2014.09.037.
References
[1] Andersen G, Vestergaard K, Ingeman-Nielsen M, Jensen TS. Incidence of central
post-stroke pain. PAIN 1995;61:18793.
[2] Attal N, Cruccu G, Baron R, Haanp M, Hansson P, Jensen TS, Nurmikko T. EFNS
guidelines on the pharmacological treatment of neuropathic pain: 2010
revision. Eur J Neurol 2010;17:e111388.
[3] Boivie J, Leijon G, Johansson I. Central post-stroke pain a study of the mechanisms
through analyses of the sensory abnormalities. PAIN 1989;37:17385.
[4] Bouhassira D, Attal N, Alchaar H, Boureau F, Brochet B, Bruxelle J, Cunin G,
Fermanian J, Ginies P, Grun-Overdyking A, Jafari-Schluep H, Lantri-Minet M,
Laurent B, Mick G, Serrie A, Valade D, Vicaut E. Comparison of pain syndromes
associated with nervous or somatic lesions and development of a new
neuropathic pain diagnostic questionnaire (DN4). PAIN 2005;114:2936.
[5] Bowsher D. Central pain: clinical and physiological characteristics. J Neurol
Neurosurg Psychiatry 1996;61:629.
[6] Cesaro P, Mann MW, Moretti JL, Defer G, Roualdes B, Nguyen JP, Degos JD.
Central pain and thalamic hyperactivity: a single photon emission
computerized tomographic study. PAIN 1991;47:32936.
[7] Choi-Kwon S, Choi JM, Kwon SU, Kang DW, Kim JS. Factors that affect the
quality of life at 3 years post-stroke. J Clin Neurol 2006;2:3441.
[8] Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain Inventory.
Ann Acad Med Singapore 1994;23:12938.
[9] Craig AD, Bushnell MC. The thermal grill illusion: unmasking the burn of cold
pain. Science 1994;265:2525.
[10] Finnerup NB, Norrbrink C, Trok K, Piehl F, Johannesen IL, Sorensen JC, Jensen
TS, Werhagen L. Phenotypes and predictors of pain following traumatic spinal
cord injury: a prospective study. J Pain 2014;15:408.
[11] Garcia-Larrea L, Convers P, Magnin M, Andre-Obadia N, Peyron R, Laurent B,
Mauguiere F. Laser-evoked potential abnormalities in central pain patients:
the inuence of spontaneous and provoked pain. Brain 2002;125:276681.
[12] Garcia-Larrea L, Maarrawi J, Peyron R, Costes N, Mertens P, Magnin M, Laurent
B. On the relation between sensory deafferentation, pain and thalamic activity
in Wallenbergs syndrome: a PET-scan study before and after motor cortex
stimulation. Eur J Pain 2006;10:67788.
[13] Hansen AP, Marcussen NS, Klit H, Andersen G, Finnerup NB, Jensen TS. Pain
following stroke: a prospective study. Eur J Pain 2012;16:112836.
[14] Head H, Holmes G. Sensory disturbances from cerebral lesions. Brain
1911;34:102254.
[15] Jensen TS, Baron R. Translation of symptoms and signs into mechanisms in
neuropathic pain. PAIN 2003;102:18.
[16] Jensen TS, Baron R, Haanpaa M, Kalso E, Loeser JD, Rice AS, Treede RD. A new
denition of neuropathic pain. PAIN 2011;152:22045.
[17] Jensen TS, Lenz FA. Central post-stroke pain: a challenge for the scientist and
the clinician. PAIN 1995;61:1614.
[18] Jensen TS, Finnerup NB. Allodynia and hyperalgesia in neuropathic pain:
clinical manifestations and mechanisms. Lancet Neurol 2014;13:92435.
[19] Klit H, Finnerup NB, Andersen G, Jensen TS. Central poststroke pain: a
population-based study. PAIN 2011;152:81824.
[20] Klit H, Finnerup NB, Jensen TS. Central post-stroke pain: clinical characteristics,
pathophysiology, and management. Lancet Neurol 2009;8:85768.
[21] Klit H, Finnerup NB, Jensen TS. Dening post-stroke pain: diagnostic
challenges. Lancet Neurol 2010;9:3445 [Authors reply].
[22] Lampl C, Yazdi K, Roper C. Amitriptyline in the prophylaxis of central
poststroke pain. Preliminary results of 39 patients in a placebo-controlled,
long-term study. Stroke 2002;33:30302.
[23] Leijon G, Boivie J, Johansson I. Central post-stroke pain neurological
symptoms and pain characteristics. PAIN 1989;36:1325.
[24] Lenz FA, Gracely RH, Baker FH, Richardson RT, Dougherty PM. Reorganization
of sensory modalities evoked by microstimulation in region of the thalamic
principal sensory nucleus in patients with pain due to nervous system injury.
J Comp Neurol 1998;399:12538.

2706

H. Klit et al. / PAIN 155 (2014) 26992706

[25] MacGowan DJ, Janal MN, Clark WC, Wharton RN, Lazar RM, Sacco RL, Mohr JP.
Central poststroke pain and Wallenbergs lateral medullary infarction: frequency,
character, and determinants in 63 patients. Neurology 1997;49:1205.
[26] Pain terms: a current list with denitions and notes on usage. In: Merskey
H, Bogduk N, editors. Classication of chronic pain. Seattle: IASP Press; 1994.
p. 20914.
[27] Naess H, Lunde L, Brogger J, Waje-Andreassen U. Post-stroke pain on long-term
follow-up: the Bergen stroke study. J Neurol 2010;257:144652.
[28] Peyron R, Garcia-Larrea L, Gregoire MC, Convers P, Lavenne F, Veyre L, Froment
JC, Mauguiere F, Michel D, Laurent B. Allodynia after lateral-medullary
(Wallenberg) infarct. A PET study. Brain 1998;121:34556.
[29] Peyron R, Garcia-Larrea L, Gregoire MC, Convers P, Richard A, Lavenne F, Barral
FG, Mauguiere F, Michel D, Laurent B. Parietal and cingulate processes in
central pain. A combined positron emission tomography (PET) and functional
magnetic resonance imaging (fMRI) study of an unusual case. Pain 2000;84:
7787.
[30] Rolke R, Baron R, Maier C, Tolle TR, Treede RD, Beyer A, Binder A, Birbaumer N,
Birklein F, Botefur IC, Braune S, Flor H, Huge V, Klug R, Landwehrmeyer GB,
Magerl W, Maihofner C, Rolko C, Schaub C, Scherens A, Sprenger T, Valet M,
Wasserka B. Quantitative sensory testing in the German Research Network on
Neuropathic Pain (DFNS): standardized protocol and reference values. Pain
2006;123:23143.

[31] Roosink M, Geurts AC, Ijzerman MJ. Dening post-stroke pain: diagnostic
challenges. Lancet Neurol 2010;9:3445.
[32] Sprenger T, Seifert CL, Valet M, Andreou AP, Foerschler A, Zimmer C, Collins DL,
Goadsby PJ, Tolle TR, Chakravarty MM. Assessing the risk of central post-stroke
pain of thalamic origin by lesion mapping. Brain 2012;135:253645.
[33] Treede RD, Jensen TS, Campbell JN, Cruccu G, Dostrovsky JO, Grifn JW,
Hansson P, Hughes R, Nurmikko T, Serra J. Neuropathic pain: redenition and
a grading system for clinical and research purposes. Neurology 2008;70:
16305.
[34] Vestergaard K, Nielsen J, Andersen G, Ingeman-Nielsen M, Arendt-Nielsen L,
Jensen TS. Sensory abnormalities in consecutive, unselected patients with
central post-stroke pain. Pain 1995;61:17786.
[35] Weimar C, Kloke M, Schlott M, Katsarava Z, Diener HC. Central poststroke
pain in a consecutive cohort of stroke patients. Cerebrovasc Dis 2002;14:
2613.
[36] Widar M, Samuelsson L, Karlsson-Tivenius S, Ahlstrom G. Long-term pain
conditions after a stroke. J Rehabil Med 2002;34:16570.
[37] Woolf CJ. Central sensitization: implications for the diagnosis and treatment of
pain. PAIN 2011;152:S2S15.
[38] Zeilig G, Enosh S, Rubin-Asher D, Lehr B, Defrin R. The nature and course of
sensory changes following spinal cord injury: predictive properties and
implications on the mechanism of central pain. Brain 2012;135:41830.