Journal of Clinical Anesthesia (2015) 27, 2332

Original Contribution

Pupil diameter during postanesthetic recovery is

not influenced by postoperative pain, but by the
intraoperative opioid treatment,
Christian Dual MD, PhD (Hospital Practitioner)a,b,c,,
Hugues Julien MB (Hospital Practitioner)d,1 ,
Bruno Pereira PhD (Biostatistician) e ,
Bertrand Abbal MB (Hospital Practitioner)d,2 ,
Charlotte Baud MB (Hospital Practitioner)d ,
Pierre Schoeffler MD (Professor of Anesthesiology)c,d,f
a

CHU Clermont-Ferrand, Centre de Pharmacologie Clinique, F-63000 Clermont-Ferrand, France

Inserm, CIC1405, F-63000 Clermont-Ferrand, France
c
Inserm, U1107 Neuro-Dol, F-63000 Clermont-Ferrand, France
d
CHU Clermont-Ferrand, Ple Anesthsie-Ranimation, Hpital Gabriel-Montpied, F-63000 Clermont-Ferrand, France
e
CHU Clermont-Ferrand, Direction de la Recherche Clinique et des Innovations, F-63000 Clermont-Ferrand, France
f
Univ Clermont1, F-63001 Clermont-Ferrand, France
b

Received 11 June 2013; revised 9 September 2014; accepted 9 September 2014

Keywords:
Pain, postoperative;
Pupillometry;
Analgesia;
General anesthesia

Abstract
Study Objective: To investigate whether pupil diameter (PD) measured during scotopic conditions is
influenced by pain in conscious patients in the early postoperative period.
Design: Prospective, observational, cohort study.
Setting: Single-center, postanesthesia care unit (PACU).
Patients: Patients scheduled for a surgery during general anesthesia.
Interventions: Baseline PD was measured the day before surgery. Patients were observed on admission
to the PACU, immediately after extubation, during the different steps of analgesic intervention (demand,
relief, plus intermediate measures when relevant), and either at discharge or 3 hours after admission.
Measurements: PD, pain (numerical rating scale), and alertness (Observer's Assessment of Alertness/
Sedation scale).
Main Results: Of 103 patients enrolled, 80 required analgesia in the PACU and completed follow-up. Pain
intensity evolved in line with expectations (temporary increase then relief), and alertness increased with time.
PD increased from low mean values at admission to the PACU (40% of baseline) to a plateau throughout the

Financial support and sponsorship: None.

Conflicts of interest: Idmed (Marseille, France) lent the Neurolight pupillometer for the purpose of the study.
Correspondence: Christian Dual, MD, PhD, Centre de Pharmacologie Clinique (Inserm CIC 501), CHU de Clermont-Ferrand, 58 Rue Montalembert,
63000 Clermont-Ferrand, France. Tel.: +33 473178418; fax: +33 473178412.
E-mail address: cduale@chu-clermontferrand.fr (C. Dual).
1
Hugues Julien is now working at the CH Saint-Pierre de la Runion (France).
2
Bertrand Abbal is now working at the CHU Montpellier (France).

http://dx.doi.org/10.1016/j.jclinane.2014.09.006
0952-8180/ 2014 Elsevier Inc. All rights reserved.

24

C. Dual et al.
rest of the study period (80% of baseline) and was not related to pain intensity. Multivariate analyses
suggested that the factors influencing PD (or its value related to baseline) were time since extubation and the
type of opioid (remifentanil, sufentanil, or sufentanil at high doses) administered during surgery.
Conclusions: Because of a residual effect of intraoperative opioids and a level of nociceptive stimulation
lower than in surgical conditions, PD is not significantly influenced by early postoperative pain or pain relief.
2014 Elsevier Inc. All rights reserved.

1. Introduction
The current pharmacological options to counteract nociception are the intraoperative administration of intravenous opioids,
followed by postoperative analgesic regimens, guided by pain
assessment tools based mostly on the patient's subjective
sensation [1]. However, during surgery receiving general
anesthesia, the challenge is to provide sufficient analgesia to
avoid central sensitization of pain, without excessive amount of
opioids and the resultant hyperalgesia [2,3].
The measurement of scotopic (ie, under low light) pupil
diameter (PD) appears as an interesting approach to estimate
nociception and analgesia in unconscious patients. Indeed,
although nociceptive stimulation increases PD in an
intensity-dependent manner [4], opioids induce a dosedependent constriction (miosis) [5]. Pupil dilation induced by a
transient nociceptive stimulus, also known as the pupil
dilation reflex (PDR), is regularly used in experimental
research protocols [5-8], whereas infrared pupillometers
dedicated to the practice of anesthesia are now available on
the European market [9,10]. To build a mathematical model
able to predict excessive nociception during surgery, the
relationship between pain and PD was studied in conscious
patients. The available data in conscious subjects had been
obtained only with mild nociceptive stimulation [4,11,12].
Feeling that such conditions were far from the clinical
context, the immediate postoperative period was studied.
Two similar observational studies failed to find a significant
relation between early postoperative pain and PD [13,14].

2. Materials and methods

Ethical approval for this study was provided by the Comit
de Protection des Personnes Sud-Est I, Saint-Etienne, France
(Chairperson Prof Ph. Rusch) on October 18, 2010. Patients
were included if aged 18-90 years and scheduled for surgery
requiring general anesthesia and opioids in the early postoperative period. Exclusion criteria were emergency surgery,
regional anesthesia, planned intensive care unit admission,
incapacity to understand the protocol and sign consent, ocular
disease, or any abnormality that would interfere with pupillometry. Patients gave signed consent the evening before surgery.
Pupillometry was performed by the same examiner for each
patient, using an infrared-based Neurolight pupillometer

(Idmed, Marseille, France). Scotopic conditions were obtained

with the device's light-tight occlusive silicone collar between the
camera and the edges of the orbit and with the examiner
shielding the contralateral eye tightly with a hand. After
application, the patient was asked to keep the eye open, and PD
was noted after stable values were obtained (30 seconds in
general). The baseline preoperative measurement of PD was
undertaken the evening before surgery. To check any
abnormality of the pupils, both eyes were controlled, and the
PD was also measured under photopic conditions (ie, elicitation
of photomotor reflex by a 320-lx flashlight applied during 1
second). All further measurements were undertaken on the right
pupil or on the left in case of abnormality of the right eye.
No measurement was performed during surgery. The time
point for the end of intraoperative analgesia was the time of
discontinuation of the infusion when remifentanil was the sole
opioid or the 20th minute after either the last bolus or the
discontinuation of the infusion, otherwise. The protocol for
anesthesia and analgesia and the decisions for other care in the
postanesthesia care unit (PACU) were left to the choice of the
practitioner in charge of the patient, but the morphine titration
protocol was standardized. Patients were transferred after
surgery to PACU, intubated, and ventilated. Tracheal extubation
was performed on recovery from anesthesia and effective
spontaneous ventilation. On extubation, vital signs were
monitored every 30 minutes until discharge. Systematic
analgesia included (except if contraindicated) intravenous (IV)
paracetamol (1 g/6 hours) and could be supplemented with
ketoprofen (50-100 mg/8 hours) or nefopam (20 mg/6 hours),
depending on the medical history. All nurses were trained in
pain assessment and able to perform morphine titration. After
return of full consciousness, patients were questioned about the
presence of pain and were shown a 0-10 visually enlarged
laminated numerical rating scale [15]. When pain score was over
3/10, IV morphine was titrated every 5 minutes by 3-mg
increments (2 mg in patients weighing b 60 kg or older than 85
years), and pain was assessed every 5 minutes until score was
b 3. If the patient was asleep after extubation and did not answer
when asked to quote pain, no further attempt was made at
arousal, and the pain score was scored as 0. When the patient
was conscious and claimed to feel strong pain but not able to
quote pain on a visual scale, due to his/her perception of
emergency to receive analgesics, 10 was scored. Titration was
stopped in case of bradypnea, hypoxemia, or excessive sedation.
In some cases, the opiate administered for titration was not
morphine, but either oxycodone or tramadol, by bolus of 2-3 mg

Pupil diameter and early postoperative pain

or 100 mg, respectively. When pain relief was obtained, patients
were provided with a patient-controlled analgesia (PCA)
morphine pump before discharge to the ward. Discharge was
allowed when the following criteria were met: Aldrete score N 9,
no nausea or vomiting, and pain score b 3/10. In certain cases
involving after major procedures performed in patients with poor
medical conditions, an overnight stay in the PACU was scheduled.
The planned observations specific to the present study were
undertaken at T0 (time of admission to PACU), Text (first time
the patient could communicate after extubation), first demand
for analgesia, when pain relief was reached, and at Tend (ie, at
discharge if the patient stayed b 3 hours in the PACU or at T0 +
3 hours otherwise). When several interventions were needed to
relieve pain, intermediate measurements were also performed.
A single observation included all of the following measurements: pain score, PD, and the Observer's Assessment of
Alertness/Sedation scale (OAA-S) [16]. Anesthetic medications administered intraoperatively were also recorded.
According to within-sample distribution or to clinical
relevance, some numerical variables were transformed into
nominal, and some nominal variables were merged to reduce the
number of modalities. Surgical procedure was classified, in
relation to experience of the level of analgesia required (level I,
unlikely to stay long in PACU and to need PCA; level II,
analgesic intervention usually required and PCA sometimes
needed; level III, analgesic intervention and PCA almost always
needed). Age was transformed into a binomial outcome (65
years or not). Each patient was also attributed a highest level of
surgery performed (none or minor, outpatient surgery, excluding
laparoscopy; major, same as level III defined above or multiple
procedures for trauma; intermediate, any other type). The type of
intraoperative opioid analgesia was defined according to the drug
administered and the distribution of the data for the administered
doses. When the opioid given in the PACU was not morphine,
the equivalent dose of IV morphine was estimated after the
following conversion: IV doses (in milligrams) of oxycodone
were multiplied by 2, according to the drug label [17], and those
of tramadol were divided by 12, on the basis of a study
comparing IV titrated morphine and tramadol in PACU [18].
Finally, the relative PD (rPD) was calculated as the ratio of each
postoperative value of PD to the baseline preoperative value.
Different outcomes were defined to indicate within
subject repeated measures. When a patient was observed in
PACU for other purposes than standard observations at T0,
Text, and T0 + 3 hours (ie, any observation before, during, or
after analgesic intervention), each of these interventions was
defined as Tn, n being the order of the intervention in time.
The delay since extubation was also calculated, as it could be
a more relevant outcome than time since the end of general
anesthesia, as it is less influenced by the kinetics of
intraoperative opioids. Finally, we defined a condition of
examination with 5 modalities, that is, preoperative
baseline, T0, Text, in PACU (all values collected in PACU
between Text and Tend pooled together), and Tend.
The quantitative data were expressed as mean SD if
normally distributed and as quartiles and range otherwise.

25
Log transformations were done if relevant. Categorical data
were expressed as number of patients/observations and
percentage of the total. For comparison between subgroups,
the Student t test (numerical Gaussian), the Mann-Whitney U
test (other numerical or ordinal), and the 2 or Fisher exact test
(categorical) were used. Multivariable analysesin which the
dependent outcome was either PD or rPD measured repeatedly
were conducted using linear mixed models, considering the
first measurement as the reference for time. In these models
with random subject effects (random intercept and slope) and
fixed effects, parameters were estimated by restricted
maximum likelihood. Multicollinearity was avoided by
reducing the number of either modalities or covariates, with
the help of descriptive and factorial analysis. Statistical
analyses were performed using Stata V10 (StataCorp, College
Station, TX, USA) and XLStat (Addinsoft, Paris, France).
Significance level (type I error) was set at 0.05 for all analyses.
Figures were generated using Microsoft Office Excel 2003.
Sample size estimation was set to evidence a correlation
between PD and pain intensity, based on data collected in
similar conditions. First, we measured PD in 28 sedated
intensive care patients (78 observations) in current conditions
of a moderately painful current nursing (turning the patient
onto the side), sedation and analgesia being slightly increased
before nursing. Mean PD went from 2.9 0.5 mm before
nursing to 3.8 1.2 mm during nursing (P b .0001). We also
used data from studies conducted in healthy volunteers during
general anesthesia, in which a relevant noxious stimulation
was applied; mean PD was decreased with opioid analgesia by
1.7-3.4 mm [5,6]. We considered that 51 observations would
be necessary to show a difference of 1 mm with the greatest SD
noted in previous data (ie, 1.4 mm [6]) with = 5%, 1 =
95%, and = 0 (cross-over design, bilateral hypothesis).
Considering that only the patients who would require analgesic
intervention would be analyzed, recruitment was extended
to 100 patients.

3. Results
The flow chart of the participants is displayed in Fig. 1.
Eighty patients (77.7% of the included) needed analgesic
intervention in the PACU and were considered for primary
analysis (main sample). Tables 1-3 describe the whole
sample and subgroups. Only the highest level of previous
surgeries was significantly greater in the main sample than in
the patients who did not need additional analgesia.
A nominal variable was built to represent the type of
intraoperative opioid analgesia. As distribution for doses of
remifentanil was Gaussian, patients who received this drug
constituted 1 modality remifentanil. As distribution for
doses of sufentanil was bimodal, patients who received
sufentanil were classified as high sufentanil or normal
sufentanil, according to a cutoff value of 0.75 g/kg. As
some patients who received remifentanil or sufentanil also

26

C. Dual et al.

Fig. 1 Flow chart of the participants. The diagram at the bottom displays the number of patients within the main sample, that is, those who
needed an analgesic intervention in PACU and were considered for the primary analysis (n = 80), depending on (i) the number of observations
undertaken in the PACU after the 2 primary observations (T0, admission to the PACU; Text, early observation after extubation) and (ii) stay
of the patient over the third hour after admission (T0 + H3).

received 1 single dose of tramadol at the end of surgery, this

variable was nested within the 3 modalities. Compared to
patients who had sufentanil, those of the remifentanil
subgroup were similar for age, gender, duration of surgery,
and need for additional analgesia. However, the type of
surgery for which anesthesia was indicated was less often
major in the remifentanil subgroup (0% vs 20%; P =
.0004). In this subgroup, the main indications were carotid
endarterectomy (n = 16) and thyroidectomy (n = 7).
The values of PD at baseline, T0, Text, and Tend are shown in
Table 4. In 1 patient whose right eye could not be examined,
the left eye was considered. Baseline values were statistically
similar to standard values reported in healthy volunteers in
similar conditions [19]. Both scotopic and photopic PDs were
well correlated side to side (R2 = 0.822 and 0.770,
respectively), as was scotopic PD to photopic PD (R2 =
0.587 for the right eye). There was no difference between
subgroups, whatever the time of measurement.
Within the main sample, 346 observations were performed
in PACU (excluding those at T0 + 3 hours), out of which 274
(79.2%) were associated with a pain score N 3/10 and were,
therefore, eligible for an analgesic intervention. Within these
observations, the median pain intensity was 5/10 (interquartile

range, 4.5-6; max, 10). The mean rate of observations with a

pain score N 3/10 was 100% for all the first observations done
after Text in PACU, and this rate significantly decreased from
T1 to T8 (P = .002). The median total dose of opioids given for
analgesia in the PACU was 7 mg of morphine or equivalent
(interquartile range, 4-11.3; range, 0-34). No serious adverse
event was noted.
Fig. 2 displays the 3 main outcomes of the study. All data
collected between Text and Tend were pooled together,
regardless of pain and analgesia. The OAA-S score increased
to reach its maximal value at Tend, but 6 patients had an OAA-S
score at 4 at Tend. Pain score increased from T0 to Text and
reached the 3/10 target at Tend, except for 4 patients who
received additional analgesia before discharge (2 with
peripheral nerve block after hip surgery and 2 with opioid).
PD decreased from baseline to T0, then tended to increase from
T0 to Tend, andcontrary to pain scoredid not peak between
Text and Tend.
Univariate analyses to identify preoperative variables
influencing baseline PD found significance for age, American
Society of Anesthesiologists (ASA) score, hypertension,
diabetes, neuroleptic medication, and opioid medication. A
multiple ANalysis Of VAriance tested the effects of all these

Pupil diameter and early postoperative pain

Table 1

27

Preoperative characteristics of the patients

Patients' characteristics
Female gender
Age (y)
Age N 65 y
Weight (kg)
ASA score
I
II
III
Previous medical history
Hypertension
Diabetes
Stroke
Coronary disease
Dyslipidemia
Hepatic insufficiency
Renal insufficiency
Neuropsychiatric disorder
Cancer
Highest level of previously performed surgeries
None or minor
Intermediate
Major
Current medications
-Blocker
Calcium-channel blocker
Antidepressant
Neuroleptic
Benzodiazepine or equivalent
Opioid

Whole sample
(n = 103)

Early observations only

(n = 23)

Main sample
(n = 80)

Between subgroups
(P value)

39 (37.9)
62.3 14.9
48 (46.6)
73.0 15.2

7 (30.4)
62.5 15.9
11 (47.8)
73.5 17.1

32 (40.0)
62.2 14.6
37 (46.3)
72.9 14.7

.555
.943
.894
.876
.983

21 (20.4)
55 (53.4)
27 (26.2)

5 (21.7)
12 (52.2)
6 (26.1)

16 (20.0)
43 (53.8)
21 (26.3)

37 (35.9)
13 (12.6)
15 (14.6)
17 (16.5)
13 (12.6)
2 (1.9)
2 (1.9)
10 (9.7)
4 (3.9)

6 (26.1)
3 (13.0)
4 (17.4)
3 (13.0)
2 (8.7)
0 (0)
0 (0)
2 (8.7)
1 (4.3)

31 (38.8)
10 (12.5)
11 (13.8)
14 (17.5)
11 (13.8)
2 (2.5)
2 (2.5)
8 (10.0)
3 (3.8)

46 (44.7)
35 (34.0)
22 (21.4)

16 (69.6)
4 (17.4)
3 (13.0)

30 (37.5)
31 (38.8)
19 (23.8)

30 (29.1)
0 (0.0)
14 (13.6)
9 (8.7)
16 (15.5)
9 (8.7)

6 (26.1)
0 (0)
1 (4.3)
1 (4.3)
4 (17.4)
3 (13.0)

24 (30.0)
0 (0.0)
13 (16.3)
8 (10.0)
12 (15.0)
6 (7.5)

NC

.024

NC

NC = not compared.
Description of the whole sample and the subgroups, as defined according to the need for analgesic intervention (see also Fig. 1). Numerical data are
expressed as mean SD or median (range). Categorical data are expressed as number of patients and percentages. Between-subgroups comparisons were
performed only if they were relevant and if variables were independent of each other (P values are given in the right column).

variables but the ASA score, plus the interactions between age
and each other variable. The model was predictive (P = .001),
diabetes being the only independent factor (mean PD, 4.2 vs 4.9
in diabetic and nondiabetic, respectively). Same analyses were
conducted with PD and rPD measured at T0 or Text as dependent
outcomes and the type of intraoperative opioid analgesia
replacing the level of the surgery in the model, as these 2
variables were collinear. None of them reached significance.
Table 5 shows the results of the linear mixed model
analyses. In analysis 1, the observations undertaken at T0 and
Text were excluded to keep conditions of good consciousness.
No significant effect was found, while the correlation between
PD/rPD and pain was poor (R2 b 0.1%). Analysis 2 checked if
adding factors related to the use of opioids (intraoperative or
postoperative) could influence the model. This was found for
the type of intraoperative opioid, only on rPD. When this factor
was kept as the sole fixed effect, the model was predictive for
PD and rPD (analysis 3). Fig. 3 shows the time course of PD
during the period of interest of these analyses, depending of the

type of intraoperative opioid analgesia, indicating that the

differences between subgroups increased with the time spent in
recovery. Analysis 4 was conducted to clarify the fact that the
principal analysis was inconclusive, while the time course of
PD displayed a V-shaped curve (Fig. 2). It confirmed that PD
stayed significantly lower in the postoperative conditions than
the preoperative condition.

4. Discussion
PD assessment in conscious postoperative patients had
already been reported, although with different aims and
conditions. In 14 patients observed after abdominal surgery,
who had analgesia by meperidine at request on the first
postoperative day, then an incremental administration of IV
alfentanil to achieve analgesia (6-8 g/kg- ), alfentanil
decreased PD from 4 to 3 mm, and a nonsignificant correlation

28
Table 2

C. Dual et al.
Intraoperative outcomes (surgery and anesthesia)

Surgical procedure
Parotidectomy
Thyroidectomy
Lumbar stenosis
Lumbar discal herniation
Total hip replacement
Carotid endarterectomy
Abdominal aortic surgery
Thoracotomy
Cystectomy
Nephrectomy
Complete prostatectomy
Level of surgery
I
II
III
Duration of surgery (min)
Premedication
Hydroxyzine
Gabapentin/pregabalin
Benzodiazepine or equivalent
Induction of general anesthesia a
Propofol
Thiopentone
Etomidate
Midazolam
Ketamine
Halogenated agent
Intravenous lidocaine
Propofol: dose for induction (mg/kg)
Maintenance of general anesthesia
Desflurane
Sevoflurane
Propofol
Propofol: mean infusion rate if target
controlled (mg/kg/h)
Intraoperative analgesia (opioids)
Remifentanil
Remifentanil + tramadol b
Sufentanil b 0.75 g/kg c
Sufentanil b 0.75 g/kg c + tramadol b
Sufentanil N 0.75 g/kg c
Sufentanil: total dose (g/kg)
Remifentanil: mean infusion
rate (g/kg/min)
Intraoperative coanalgesia (nonopioids)
Paracetamol
Nefopam
Ketoprofen
Steroid
Single-dose local anesthesia
Atropine (0.5-1 mg as single dose)
Ephedrine
Ephedrine: total dose (mg)
Droperidol (1.25 mg as single dose)

Whole sample
(n = 103)

Early observations only Main sample

(n = 23)
(n = 80)

1
10
11
21
26
17
1
6
4
2
3

0
2
3
5
5
3
1
1
0
0
2

11 (10.7)
76 (73.8)
16 (15.5)
100 [70-150] (25-430)

2 (8.7)
17 (73.9)
4 (17.4)
95 [65-143] (35-270)

75 (72.8)
24 (23.3)
6 (5.8)

17 (73.9)
7 (13.4)
2 (8.7)

101 (98.1)
1 (1.0)
2 (1.9)
32 (31.1)
11 (10.7)
0 (0)
5 (4.9)
2.5 1.1

23 (100)
0 (0)
0 (0)
7 (30.4)
2 (8.7)
0 (0)
2 (8.7)
2.5 0.8

12 (11.7)
65 (63.1)
28 (27.2)
0.16 0.07

3 (13.0)
15 (65.2)
5 (21.7)
0.18 0.11

Between subgroups
(P value)
NC

1
8
8
16
21
14
0
5
4
2
1
.916
9 (11.3)
59 (73.8)
12 (15.0)
105 [70-150] (25-430) .430
NC
58 (72.5)
17 (21.3)
4 (5.0)
NC
78 (97.5)
1 (1.3)
2 (2.5)
25 (31.3)
9 (11.3)
0 (0)
3 (3.8)
2.5 1.2
NC
.898
9 (11.3)
50 (62.5)
23 (28.8)
0.15 0.06
NC
.875 d

23 (22.3)
1 (1.0)
41 (39.8)
23 (22.3)
15 (14.6)
0.66 0.44
0.11 0.05

4 (17.4)
0 (0)
9 (39.1)
6 (26.1)
4 (17.4)
0.58 0.27
0.14 0.08

19 (23.8)
1 (1.3)
32 (40)
17 (21.3)
11 (13.8)
0.69 0.48
0.10 0.04

102 (99.0)
70 (68.0)
28 (27.2)
8 (7.8)
1 (1.0)
4 (3.9)
50 (48.5)
13.0 9.0
8 (7.8)

23 (100)
16 (69.6)
5 (21.7)
2 (8.7)
0 (0)
NA
13 (56.5)
15.5 8.4
2 (8.7)

79 (98.8)
54 (67.5)
23 (28.8)
6 (7.5)
1 (1.3)
4 (5)
37 (46.3)
12.2 9.2
6 (7.5)

NC

NC

NC

Pupil diameter and early postoperative pain

Table 3

29

Flow of the patients in PACU

Time between end of analgesia a

and admission in PACU (min)
Time between extubation and first
postextubation observation
Text (min)
Time between end of analgesia a
and extubation (min)
Patients still present in PACU
at T0 + 3 h
Time between end of analgesia a
and discharge from PACU (min) b
Actual time spent in PACU (min) b

Whole sample
(n = 103)

Early observations
only (n = 23)

Main sample
(n = 80)

Between subgroups
(P value)

25 [15-44] (0-105)

25 [18-40] (0-90)

25 [15-45] (0-105)

.668

10 [5-10] (5-20)

10 [10-10] (5-20)

10 [5-10] (5-20)

.283

50 [30-70] (5-305)

55 [43-75] (15-130)

50 [30-66] (5-305)

.224

29 (28.2)

5 (21.7)

24 (30.0)

1.000

180 [160-200] (100-425) 165 [140-184] (110-220) 190 [160-200] (100-425) NC

145 [115-167] (60-335)

133 [115-154] (60-170)

150 [119-170] (75-335)

NC

Description of the whole sample and the subgroups, as defined according to the need for analgesic intervention (see also Fig. 1). Numerical data are
expressed as mean SD or median [interquartile range] (range). Categorical data are expressed as number of patients and percentages. Between-subgroups
comparisons were performed only if they were relevant and if variables were independent of each other (P values are given in the right column).
a
See Materials and methods for definition.
b
Only for the patients discharged before T0 + 3 hours.

Table 4

General aspects of pupil diameter

Preoperative scotopic PD (baseline)

Scotopic
Photopic
Scotopic PD: measurements in the PACU
At arrival in the PACU (T0)
PD
rPD
Just after extubation (Text)
PD
rPD
At the end of the study (Tend) b
PD
rPD

Whole sample
(n = 103)

Early observations
only (n = 23)

Main sample
(n = 80)

Between subgroups
(P value)

4.8 0.9
3.3 1.1

4.7 1.0
3.4 1.2

4.8 0.8
3.2 1.1

.544
.611

1.9 (1.6-2.4)
0.40 (0.36-0.50)

1.9 (1.5-2.3)
0.39 (0.36-0.44)

1.9 (1.6-2.6)
0.40 (0.36-0.54)

.140 a
.195 a

3.2 (2.2-4.0)
0.68 (0.53-0.85)

2.8 (2.5-3.4)
0.58 (0.52-0.79)

3.3 (2.6-4.1)
0.72 (0.55-0.82)

.140 a
.255 a

3.8 1.1
0.79 0.19

3.6 0.9
0.80 0.20

3.8 1.1
0.79 0.19

.492
.868

Scotopic PD measured at the right eye at the times of observation that were standardized for all patients, within the whole sample and the subgroups, as
defined according to the need for analgesic intervention (see also Fig. 1). Numerical data are expressed as mean SD or median (interquartile range). The
rPD is the ratio to the baseline preoperative value.
a
The Student t test performed on log-transformed values.
b
Either at discharge from PACU or at T0 + 3 hours.

(r = 0.34) was noted between the analgesic and the miotic

potency of the drug [20]. Two recent studies were conducted in
a setting close to ours, after various surgeries performed during
balanced general anesthesia. Aissou et al [13] observed 61
patients in whom a PDR was elicited by an intentional pressure
applied to the wound and measured before and after morphine

given for pain relief. Although neither spontaneous pain before

morphine nor pain relief after morphine influenced the baseline
PD before the additional noxious stimulation, both variables
were well correlated to the intensity of PDR. Kantor et al [14]
also measured PD 145 postoperative patientsof which 20%
needed morphine for pain reliefwithout inducing any PDR.

Description of the whole sample and the subgroups, as defined according to the need for analgesic intervention (see also Fig. 1). Numerical data are
expressed as mean SD or median [interquartile range] (range). Categorical data are expressed as number of patients and percentages. Between-subgroups
comparisons were performed only if they were relevant and if variables were independent of each other (P values are given in the right column).
a
Possible combination of 2 drugs for induction, that is, the sum of rows may exceed 100%.
b
Fifty to one hundred milligrams given as a single dose at the end of surgery.
c
The cutoff value (0.75 g/kg) has been defined according to the distribution, see Results section for details.
d
For the test, the variable tramadol was nested within the 3 main modalities.

30

Fig. 2 Time course of the outcomes. Time course of the main

outcomes registered during the study, that is, the OAA-S score; the pain
score on a numerical rating scale from 0 to 10; and the scotopic PD
measured at the right eye of the patient. For each outcome, the diagrams
display the median value (white circles), the first and third quartile
(dashes), and the extreme values (crosses) at each period of
measurement. The 5 points of measurement were baseline preoperative
values (Pre-op); on arrival at the PACU (T0); just after extubation
(Text); during the stay in PACU (PACU, pooled data from all the
measured values); end of the observation study (Tend), that is, either at
discharge for patients who stayed less than 3 hours in the PACU or at T0
+ 3 hours otherwise.

They found neither relation between pain and PD nor relation

between pain relief and PD changes. The early postoperative
setting may not be suitable to show a relationship between PD
and spontaneous pain. An effect on PD can be revealed only if
pain is reinforced by stimulation of the wound (a situation closer
to intraoperative than to postoperative conditions) ormildly
if measurements are made at distance from anesthesia. This is
also consistent with data from studies in healthy volunteers, in
which strong electrical stimulations applied during general
opioidless anesthesia sensibly increased PD [5-8], whereas
milder stimulations applied in conscious subjects induced only a
small increase (b 0.5 mm) [4,11,12].

C. Dual et al.
As the mean PD throughout the stay in the PACU did not
exceed 80% of the preoperative values, we suspected a residual
effect of effects of general anesthesia. This was supported by
the significant effect of the type of intraoperative opioid, in
accordance with the elimination half-life of the drugs. Most of
our patients who received intraoperative remifentanil did not
receive a systematic compensating bolus of morphineas
often practicedbecause we currently use this drug for carotid
or thyroid surgery (here considered as level I). In nonpainful
conditions, the -agonist opioids have a direct miotic effect,
even at subanesthetic doses [5,21-23]. This direct action is
attributed to a blockade of inhibition of the Edinger-Westphal
nucleus neurons [8]. In nociceptive conditions, this direct
action is concomitant to the reduction of pain-induced pupil
dilation (PDR) through analgesia, a phenomenon evidenced in
subjects under general anesthesia [5-8]. In addition, the
opioid-induced myosis is dose dependentwhatever the
nociceptive stimulationwith a floor effect slightly inferior
to 2 mm [5].
As in the 2 previous studies conducted in PACU, we
intentionally included a heterogeneous population and had a
liberal attitude toward the practice of anesthesia and analgesia,
aiming to reflect routine clinical practice [13,14]. In such
observational design, it is difficult to distinguish the effects of
pain (and analgesia in general) and those of opioids
administered postoperatively. Nevertheless, our multivariable
analysis did not find any effect of the dose of opioid
administered before observation, whereas Kantor et al noted
in 27 patients who received morphine titration a mild but
significant reduction of PD by morphine (from 3.8 to 3.4 mm)
[14]. Lower doses of morphine were administered in our study
(7 vs 9 mg). The main information is the observed discrepancy
between the clinical effectiveness of postoperative opioids and
their weak effect on PD (when measured without additional
stimulation). Besides a masking effect of intraoperative opioids
on PD, it cannot be excluded that a part of postoperative
analgesia was a placebo effect, as already evidenced in a similar
setting [24]. In addition, the doses administered postoperatively
are low for safety reasons and then may not block nociceptive
pathways. For example, 1 mg/kg of morphine given in rats acts
on supraspinal aspects of pain perception [25], whereas a 6-fold
superior dose is needed to impair the response of spinal
nociceptive neurons [26].
The current study has some methodological limitations,
mostly due to its observational design. First, the accuracy of pain
scales may be impaired in elderly patients, leading to
underestimation in this population [27]. Second, pupil dilation
can also be elicited by nonpainful conditions, such as mental
tasks [28], anxiety [29], or even spontaneous awakeness [30].
However, PD changes in such conditions are in a much smaller
range than those observed in nociceptive PDR; furthermore,
mental aspects may also influence pain intensity [1]. Third,
another bias may be due to heterogeneity of the sample, in which
small subgroups of patients may have an impaired pupil function.
This is illustrated by the effect of diabetes that we observed and
which had already been reported [31], but impairment may also

Pupil diameter and early postoperative pain

Table 5

31

Factors influencing pupil diameter in PACU

Analysis 1 Analysis 2
(principal)

Sample for analysis

Within-subject repetition
Fixed effects
Pain score
Dose of opioid before observation
Type of intraoperative opioid
Dependent outcome: PD
P value (model)
Post hoc analyses

Main sample
DSE
Tn

Tested
Not tested

Tested

Analysis 3

Analysis 4

DSE

Whole sample
Condition of
examination

Not tested

Not tested

Tested
PD rPD PD rPD
PD
rPD
PD
rPD
.293 .216 .183 .003
.017
.0002
b .0001 b .0001
ND
ND Rf nS and hS DSE (P = .017); Rf hS Rf nS and hS Baseline all
postoperative
conditions

Tn = time of observation in PACU; DSE = delay since extubation; ND = not done; Rf = intraoperative remifentanil; nS = intraoperative sufentanil, normal
doses; hS = intraoperative sufentanil, high doses.
Multivariable analyses using a linear mixed model. Post hoc analyses (Tukey-Kramer's test) were done only when the model was found predictive (ie, P
value for the model b .05). The rPD is the ratio to the baseline preoperative value. The variables Tn, DSE, and Condition of examination are defined in
the Materials and methods section. The modalities Rf, nS, and hS (type of intraoperative opioid) are defined in the Results section.

be due to ongoing neuroleptic medication [32]. A last concern is

the intraoperative administration of drugs likely to interact with
PD. Indeed, ephedrine and atropine are mydriatic agents [33,34]
and are a part of the current practice; they, however, have a short

action. Similarly, the neuroleptic agent droperidol may be

administered to prevent nausea but at very low doses (1.25 mg).
Atropine and droperidol were rarely administered in the current
study (b 5%), whereas in the study of Kantor et al [14], they was
given in 56% and 73% of the patients, respectively.
The early postoperative conditions were not suitable for the
identification of standard values of PD that may predict
insufficient or excessive analgesia. The lack of relation between
PD and postoperative pain may be due to (i) levels of
nociception lower than those observed during surgery or when
PDR is elicited without high opioid analgesia, (ii) a residual
influence of intraoperative opioids strong enough to mask the
effects of postoperative pain or opioids, and (iii) the interaction
of postoperative analgesia with subtle aspects of cognition
(including the placebo effect) inactive on pupil function.

Acknowledgments
Assistance with the article:
Fig. 3 Intraoperative opioids and postoperative PD. Time course of
the scotopic PD measured at the right eye of the patient, depending on
the type of opioid analgesia administered during surgery. The diagram
displays the grand mean value for all observations and the confidence
interval limits (95%) limits. The 4 points of measurement were on
arrival at the PACU (T0); just after extubation (Text); during the
stay in PACU (PACU, pooled data from all the measured values);
end of the observation study (Tend), that is, either at discharge for
patients who stayed less than 3 hours in the PACU or at T0 + 3 hours
otherwise. The subgroup of patients who received remifentanil is
represented by white circles and a dotted line; those who received
sufentanil at normal doses are represented by light gray circles and a
dashed line; those who received sufentanil at high doses are
represented by dark gray circles and a full line (see Results section
for definition of normal and high sufentanil subgroups).

- Frdric Bernert (Idmed, Marseille, France) for committing

of his society in the research.
- Fabienne Commun, Laurence Escaravage, Michal Faure,
Dominique Groslier, and Hammou Taheri (all MB and
hospital practitioners, Ple Anesthsie-Ranimation,
Hpital Gabriel-Montpied, CHU Clermont-Ferrand,
France) for overseeing the investigations.
- Denise Faulks (BDS, PhD, hospital practitioner, Odontologie,
CHU Clermont-Ferrand, France) for the correction of the
English manuscript.
- Jean-Michel Constantin (MD, PhD, Ranimation Adultes
Estaing, PRI, CHU Clermont-Ferrand, France) for providing
the tool to assess pain (visually enlarged laminated numerical
rating scale).

32

C. Dual et al.

References
[1] Breivik H, Borchgrevink PC, Allen SM, et al. Assessment of pain. Br J
Anaesth 2008;101:17-24.
[2] Wilder-Smith OH, Arendt-Nielsen L. Postoperative hyperalgesia: its
clinical importance and relevance. Anesthesiology 2006;104:601-7.
[3] Chu LF, Angst MS, Clark D. Opioid-induced hyperalgesia in humans:
molecular mechanisms and clinical considerations. Clin J Pain 2008;
24:479-96.
[4] Chapman CR, Oka S, Bradshaw DH, Jacobson RC, Donaldson GW.
Phasic pupil dilation response to noxious stimulation in normal
volunteers: relationship to brain evoked potentials and pain report.
Psychophysiology 1999;36:44-52.
[5] Barvais L, Engelman E, Eba JM, Coussaert E, Cantraine F, Kenny GN.
Effect site concentrations of remifentanil and pupil response to noxious
stimulation. Br J Anaesth 2003;91:347-52.
[6] Larson MD, Kurz A, Sessler DI, Dechert M, Bjorksten AR, Tayefeh F.
Alfentanil blocks reflex pupillary dilation in response to noxious
stimulation but does not diminish the light reflex. Anesthesiology
1997;87:849-55.
[7] Larson MD, Berry PD, May J, Bjorksten A, Sessler DI. Latency of
pupillary reflex dilation during general anesthesia. J Appl Physiol
2004;97:725-30.
[8] Larson MD. Mechanism of opioid-induced pupillary effects. Clin
Neurophysiol 2008;119:1358-64.
[9] IdMed. Analgesia. http://www.idmed.fr/pupillometrie-analgesie-105Analgesia.html. [Last access 27 June 2014].
[10] Synapsis. Video-AlgesiGraph. http://www.synapsys.fr/en/p-videoalgesigraph-vag-62.htm. [Last access 27 June 2014].
[11] Oka S, Chapman CR, Kim B, Nakajima I, Shimizu O, Oi Y. Pupil
dilation response to noxious stimulation: effect of varying nitrous
oxide concentration. Clin Neurophysiol 2007;118:2016-24.
[12] Yang LL, Niemann CU, Larson MD. Mechanism of pupillary reflex
dilation in awake volunteers and in organ donors. Anesthesiology
2003;99:1281-6.
[13] Aissou M, Snauwaert A, Dupuis C, Atchabahian A, Aubrun F,
Beaussier M. Objective assessment of the immediate postoperative
analgesia using pupillary reflex measurement: a prospective and
observational study. Anesthesiology 2012;116:1006-12.
[14] Kantor E, Montravers P, Longrois D, Guglielminotti J. Pain
assessment in the postanaesthesia care unit using pupillometry: a
cross-sectional study after standard anaesthetic care. Eur J Anaesthesiol 2014;31:91-7.
[15] Chanques G, Viel E, Constantin JM, et al. The measurement of pain in
intensive care unit: comparison of 5 self-report intensity scales. Pain
2010;151:711-21.
[16] Chernik DA, Gillings D, Laine H, et al. Validity and reliability of the
Observer's Assessment of Alertness/Sedation Scale: study with
intravenous midazolam. J Clin Psychopharmacol 1990;10:244-51.
[17] Agence Nationale de Scurit du Mdicament et des produits de sant.
OXYNORM, Rsum des caractristiques du produit. http://agence-

[18]

[19]

[20]
[21]

[22]

[23]

[24]

[25]

[26]
[27]
[28]

[29]
[30]
[31]
[32]
[33]

[34]

prd.ansm.sante.fr/php/ecodex/frames.php?specid=60209645&typedoc=R&ref=R0176152.htm. [Last access 27 June 2014].

Stamer UM, Maier C, Grond S, Veh-Schmidt B, Klaschik E, Lehmann
KA. Tramadol in the management of post-operative pain: a doubleblind, placebo- and active drug-controlled study. Eur J Anaesthesiol
1997;14:646-54.
Mantry S, Banerjee S, Naroo S, Shah S. Scotopic measurement of
normal pupil size with the Colvard pupillometer and the Nidek autorefractor. Cont Lens Anterior Eye 2005;28:53-6.
Tverskoy M, Oren M, Dashkovsky I, Kissin I. Alfentanil dose-response
relationships for relief of postoperative pain. Anesth Analg 1996;83:387-93.
Knaggs RD, Crighton IM, Cobby TF, Fletcher AJ, Hobbs GJ. The
pupillary effects of intravenous morphine, codeine, and tramadol in
volunteers. Anesth Analg 2004;99:108-12.
Fliegert F, Kurth B, Gohler K. The effects of tramadol on static and
dynamic pupillometry in healthy subjectsthe relationship between
pharmacodynamics, pharmacokinetics and CYP2D6 metaboliser
status. Eur J Clin Pharmacol 2005;61:257-66.
Verster JC, Veldhuijzen DS, Volkerts ER. Effects of an opioid
(oxycodone/paracetamol) and an NSAID (bromfenac) on driving
ability, memory functioning, psychomotor performance, pupil size,
and mood. Clin J Pain 2006;22:499-504.
Amanzio M, Pollo A, Maggi G, Benedetti F. Response variability to
analgesics: a role for non-specific activation of endogenous opioids.
Pain 2001;90:205-15.
Huang GF, Besson JM, Bernard JF. Morphine depresses the
transmission of noxious messages in the spino(trigemino)-pontoamygdaloid pathway. Eur J Pharmacol 1993;230:279-84.
Dual C, Raboisson P, Molat JL, Dallel R. Systemic morphine reduces the
wind-up of trigeminal nociceptive neurons. Neuroreport 2001;12:2091-6.
Gagliese L. Pain and aging: the emergence of a new subfield of pain
research. J Pain 2009;10:343-53.
Satterthwaite TD, Green L, Myerson J, Parker J, Ramaratnam M,
Buckner RL. Dissociable but inter-related systems of cognitive control
and reward during decision making: evidence from pupillometry and
event-related fMRI. Neuroimage 2007;37:1017-31.
Nagai M, Wada M, Sunaga N. Trait anxiety affects the pupillary light
reflex in college students. Neurosci Lett 2002;328:68-70.
McLaren JW, Hauri PJ, Lin SC, Harris CD. Pupillometry in clinically
sleepy patients. Sleep Med 2002;3:347-52.
Cahill M, Eustace P, de Jesus V. Pupillary autonomic denervation with
increasing duration of diabetes mellitus. Br J Ophthalmol 2001;85:1225-30.
Larson MD. The effect of antiemetics on pupillary reflex dilation
during epidural/general anesthesia. Anesth Analg 2003;97:1652-6.
Larson MD, Sessler DI, Washington DE, Merrifield BR, Hynson JA,
McGuire J. Pupillary response to noxious stimulation during isoflurane
and propofol anesthesia. Anesth Analg 1993;76:1072-8.
Daniel M, Larson MD, Eger EI, Noorani M, Weiskopf RB. Fentanyl,
clonidine, and repeated increases in desflurane concentration, but not
nitrous oxide or esmolol, block the transient mydriasis caused by rapid
increases in desflurane concentration. Anesth Analg 1995;81:372-8.