Acetaminophen Decreases Early Post-Thoracotomy Ipsilateral Shoulder
Pain in Patients With Thoracic Epidural Analgesia:
A Double-Blind Placebo-Controlled Study
Thien Bich Mac, MD,* François Girard, MD, FRCPC,* Philippe Chouinard, MD, FRCPC,*
Daniel Boudreault, MD, FRCPC,* Edwin R. Lafontaine, MD, FRCSC,† Monique Ruel, RN, CCRP,*
and Pasquale Ferraro, MD, FRCSC†
Objective: Despite effective epidural analgesia, up to 85%
of post-thoracotomy patients complain of moderate-to-se-
vere ipsilateral shoulder pain. This study assessed the effi-
cacy of acetaminophen in decreasing postoperative shoul-
der pain after a thoracotomy.
Design: Double-blind randomized and placebo-controlled
study.
Setting: University medical center.
Participants: 65 patients.
Intervention: Patients were randomized into 2 groups; 31
patients received acetaminophen (group A), and 34 patients
received a placebo (group P). After induction of anesthesia,
patients received either a loading dose of acetaminophen,
1000 mg intrarectally, or a placebo suppository. Thereafter,
acetaminophen, 650 mg, or a placebo, was administered
intrarectally every 4 hours for 48 hours postoperatively.
Measurements and main results: Postoperative pain at the
surgical site and shoulder pain were assessed separately
T HORACIC EPIDURAL ANALGESIA has become a stan-
dard of care for postoperative pain management after a
thoracotomy.1 In contrast to the excellent pain relief at the level
of the surgical incision, up to 85% of post-thoracotomy patients
under epidural analgesia complain of moderate-to-severe ipsi-
lateral shoulder pain.2-4
Postoperative shoulder pain following a thoracotomy is usu-
ally described as a dull and lancinating pain of moderate
intensity, usually located in the supraspinatous muscle right
above the mid-point of the scapula on the ipsilateral side from
the surgical incision. Few prospective and controlled studies
have documented the prevalence and severity of post-thoracot-
omy shoulder pain (PTSP) or found an effective treatment for
this problem. Treatment modalities investigated so far have
proven to be either inefficient, limited in their effective duration
of action, or associated with potentially deleterious side effects.
The reported prevalence of PTSP ranges from 42% to 85%.2-4
In a series of cases, Burgess et al2 suggested that ketorolac,
a nonsteroidal anti-inflammatory drug (NSAID), given regu-
larly in the postoperative period, was effective in treating PTSP
in a small subset of patients. NSAIDs are, however, associated
with important side effects and are not an ideal choice for the
old and sometimes debilitated cancer patients who frequently
make up the thoracic surgery population. Namely, the combi-
nation of postoperative fluid restriction and administration of
NSAIDs is known to be an important triggering factor for acute
renal failure.5
Acetaminophen, on the other hand, represents a safe alter-
native to NSAIDs. The authors designed this prospective,
double-blind, randomized, and placebo-controlled study to test
the hypothesis that the use of acetaminophen would reduce the
prevalence and severity of post-thoracotomy shoulder pain in
patients undergoing thoracic epidural analgesia in the first 48
hours following the surgery.
This hypothesis was based on the clinical observation that
Journal of Cardiothoracic and Vascular Anesthesia, Vol 19, No 4 (August), 2005: pp 475-478
every 4 hours for 48 hours using a numerical rating scale
(NRS). Rescue analgesia for severe shoulder pain (NRS > 7)
consisted of subcutaneous hydromorphone. Sixty-three pa-
tients experienced shoulder pain (97% prevalence). Demo-
graphic and intraoperative data were similar between the 2
groups. Average NRS for shoulder pain was higher in group
P compared with group A at 8, 12, and 16 hours postopera-
tively (3.1 ⴞ 2.9, 2.6 ⴞ 2.6, 2.3 ⴞ 2.4 vs 1.8 ⴞ 2.6, 1.2 ⴞ 1.5, 1.3
ⴞ 1.8; P < 0.05). The total dose of hydromorphone did not
differ between the 2 groups at 16, 24, and 48 hours.
Conclusion: Acetaminophen decreases post-thoracotomy
ipsilateral shoulder pain when given preemptively and reg-
ularly during the first 48 hours postoperatively in patients
who received thoracic epidural analgesia.
© 2005 Elsevier Inc. All rights reserved.
KEY WORDS: anesthesia, surgery, thoracotomy, epidural an-
algesia, shoulder pain, acetaminophen
adding acetaminophen to the postoperative analgesia regimen
seemed to greatly reduce post-thoracotomy shoulder pain in
these patients, without the side effects associated with the use
of NSAIDs.
METHODS
After institutional review board approval and signed patient in-
formed consent, 70 patients aged 18 to 80 years scheduled for a
thoracotomy in the lateral decubitus position were enrolled. They were
randomly divided into 2 groups to receive acetaminophen (group A) or
a placebo (group P) from anesthesia induction until 48 hours postop-
eratively. The exclusion criteria were: inability to understand a numer-
ical rating scale for pain assessment; proved or suspected allergy to
local anesthetics or to acetaminophen; presence of preoperative shoul-
der pain; use of narcotics, NSAIDs, or antidepressants for more than 1
week preoperatively; and contraindication to the use or placement of an
epidural catheter or to the intrarectal administration of medication.
After the placement of standard noninvasive monitors, a thoracic
epidural catheter was inserted between the T6-8 intervertebral space
before the induction of general anesthesia. The correct position of the
catheter was tested using 5 mL of 1.5% lidocaine with epinephrine
1:200,000. If a bilateral decrease of sensation to ice on more than 2
thoracic dermatomes did not occur within 15 minutes of the 1.5%
lidocaine injection, the catheter was tested again with 5 mL of 2%
carbonated lidocaine. If no decrease of sensation to ice was noted after
an additional 10 minutes of observation, then the epidural catheter was
From the *Department of Anesthesiology and †Thoracic Surgery
Division, CHUM, Hôpital Notre-Dame, Montreal, Canada.
Address reprint requests to François Girard, MD, FRCPC, Depart-
ment of Anesthesiology, CHUM, Hôpital Notre-Dame, 1560 Sher-
brooke East, Montreal, Quebec H2L 4M1, Canada. E-mail: francois.
girard.chum@ssss.gouv.qc.ca
© 2005 Elsevier Inc. All rights reserved.
1053-0770/05/1904-0010$30.00/0
doi:10.1053/j.jvca.2004.11.041
475
476
Table 1. Demographic and Surgical Data
Group P Group A
(Placebo) (Acetaminophen)
No. of patients 34 31
Male-female ratio 20:14 14:17
Age (y)* 61 ⫾ 9 60 ⫾ 14
Weight (kg)* 73.0 ⫾ 15.4 71.3 ⫾ 14.6
Duration of surgery (min)* 111.1 ⫾ 47.5 91.3 ⫾ 26.3
Duration of hospitalization (d)* 6.3 ⫾ 2.2 7.3 ⫾ 3.2
Type of surgery
Pneumonectomy 4 5
Lobectomy 17 14
Bronchogenic cyst removal 1 0
Exploratory thoracotomy 0 1
Pleurodese/talc 0 1
Wedge/biopsy/segmentectomy 12 10
*Data represent mean ⫾ SD.
reinserted and the position verified once again. The epidural solution
(bupivacaine 0.1% and fentanyl 2 ␮g/mL) was infused at a rate of 0.1
mL/kg/h intraoperatively. The infusion rate was adjusted between 4 to
16 mL/h at the discretion of the attending anesthesiologist, and boluses
of 0.1 mL/kg of the anesthetic solution were also allowed as needed.
Postoperatively, the infusion rate was adjusted as per protocol to
maintain a pain score ⬍3 at the surgical site.
After the insertion of the epidural catheter and preoxygenation,
anesthesia was induced with propofol (1-2 mg/kg) and fentanyl (1-4
␮g/kg) or sufentanil (0.2-0.5 ␮g/kg), and tracheal intubation was fa-
cilitated with pancuronium. Anesthetic maintenance consisted of
sevoflurane or desflurane, supplemented with boluses of fentanyl or
sufentanil, at the discretion of the attending anesthesiologist to maintain
blood pressure and heart rate within 20% of the patient’s baseline
value. Nitrous oxide was not used.
After anesthesia induction and positioning of the patient in the lateral
decubitus position, patients in group A received an induction dose of
acetaminophen, 1000 mg intrarectally. Then acetaminophen, 650 mg
intrarectally, was administered every 4 hours for 48 hours. In group P,
patients received placebo suppositories at exactly the same time peri-
ods. The shape, size, and appearance of acetaminophen and placebo
suppositories were identical. Each suppository was individually
wrapped with identical foil. The hospital research pharmacy provided
all the suppositories for the same patient on the morning of random-
ization. During the surgery, the ipsilateral arm was systematically
positioned the same way for all patients, that is, on an arm rest, with
about 90 degrees of anterior flexion, no abduction, and the elbow flexed
at about 45 degrees.
The severity of pain at the surgical site and the severity of the
shoulder pain were assessed at rest every 4 hours for 48 hours, starting
after the insertion of the loading dose of acetaminophen or its equiv-
alent placebo suppository. The severity of pain was assessed verbally
using a numerical rating scale (NRS), where 0 represented no pain at all
and 10 the worst possible pain. The intensity of pain at the surgical
incision and at the shoulder were assessed separately. At any time, if a
patient complained of moderate-to-severe pain (NRS ⬎ 3) in der-
matomes related to the surgical incision, 0.1 mL/kg of the epidural
solution was administered, and the infusion rate was increased by 2
mL/h as needed, to a maximum infusion rate of 16 mL/h. If pain
persisted despite these changes, the epidural infusion was then changed
for a solution containing bupivacaine 0.125% and fentanyl 2 ␮g/ml,
which was restarted at a rate of 10 mL/h following the same protocol.
In case of severe shoulder pain (NRS ⬎ 7), a rescue medication
(subcutaneous hydromorphone, 1-2 mg) was given after the assessment
of the patient by one of the investigators.
MAC ET AL
Data were stored in an Excel spreadsheet database. The results are
expressed as the mean ⫾ 1 SD except when stated otherwise. Differ-
ences in demographic and intraoperative data between the 2 groups
were sought by using the chi-squared test and the unpaired Student t
test for nonparametric and parametric variables, respectively. NRS
scores were evaluated by summarizing the data for each group and
calculating the area under the curve using the trapezium rule.6 Differ-
ences between the 2 groups were then sought at 16, 24, and 48 hours
using analysis of variance (ANOVA).7 Then differences in NRS scor-
ing between the 2 groups at each time interval were examined using the
unpaired Student t test. ANOVA for repeated measures was also used
to detect a variation in time of the NRS scoring. A P value ⬎ 0.05 was
considered significant.
RESULTS
Seventy patients were enrolled in this study, 5 of whom were
excluded after the beginning of the study. One patient withdrew
his consent 12 hours postoperatively, 2 were excluded because
the epidural catheter was not functioning properly in the recov-
ery room, 1 patient was excluded because intraoperative hypo-
tension did not allow the attending anesthesiologist to begin the
epidural infusion, and the fifth patient was excluded because all
of the NRS data were lost in the postoperative period.
Sixty five patients remained in the study, 31 in group A and
34 in group P. The demographic and surgical data for the
patients of both groups were comparable (Table 1). Thirty
patients in group A and 33 patients in group P experienced
ipsilateral shoulder pain, yielding a total prevalence of 97%.
There were no statistically significant differences between the 2
groups in the type of surgery performed, the total amount of
epidural solution administered, or the total dose of hydromor-
phone used as rescue medication at 16, 24, and 48 hours (Table
2). Summarized NRS values at each time interval are provided
in Fig 1 for shoulder pain and in Table 3 for pain at the surgical
incision. No difference was found between the 2 groups for
pain at the surgical incision. There was, however, a difference
between the NRS scores for shoulder pain between the 2 groups
for the first 16 hours of the study (area under the curve, 23.4 ⫾
21.0 for acetaminophen and 34.3 ⫾ 24.1 for placebo; P ⫽
0.05), with lower NRS scores in the acetaminophen group at 8,
12, and 16 hours postoperatively (P ⬍ 0.05). The authors
Table 2. Postoperative Data
Group P Group A
(Placebo) (Acetaminophen)
Total dose of epidural solution
administered (mL)
Bupivacaine 0.1% with
fentanyl 2 ␮g/mL* 426.0 ⫾ 210 464.7 ⫾ 172.0
Bupivacaine 0.125% with
fentanyl 2 ␮g/mL* 139.0 ⫾ 229 69.8 ⫾ 63.9
No. of patients who received
rescue medication
(hydromorphone) 15 14
Total dose of hydromorphone
(mg)
Median (95% confidence
interval) 2 (1.6-4.2) 1.25 (0.7-4.7)
*Data represent mean ⫾ SD.
ACETAMINOPHEN DECREASES POSTOPERATIVE SHOULDER PAIN AFTER THORACOTOMY 477

10

6
N.R.S. *
*
5

4
*
3

Fig 1. Severity of post-thora-

cotomy shoulder pain. NRS, nu- 2
merical rating scale (where 0 rep-
resents no pain and 10 the worst
1
possible pain); black bars, pla-
cebo group; white bars, acet-
aminophen group. Data repre- 0
4h 8h 12h 16h 20h 24h 48h
sent mean ⴞ SD; P < 0.05.

noticed that the severity of the shoulder pain was not constant of how the ketorolac-treated patients were selected among the
over time and tended to decrease toward the end of the study entire group. In addition, as shown in the present study, this
period (P ⬍ 0.05). finding could represent the natural history of the shoulder pain
and its tendency to decrease after 24 hours, as the authors failed
DISCUSSION to report the moment when the shoulder pain disappeared
This prospective, randomized, double-blind, placebo-con- completely. Patients were evaluated immediately before ke-
trolled study assessed the efficacy of acetaminophen to de- torolac administration and at 1 hour and 24 hours after its
crease postoperative ipsilateral shoulder pain after a thoracot- administration. NSAIDs, however, although possibly effective,
omy. The results obtained show that acetaminophen have the disadvantages of platelet inhibition, potential nephro-
administered preemptively and regularly in the postoperative toxicity, and variable clearance, especially in elderly patients.
period significantly decreases early postoperative ipsilateral In this study, acetaminophen was used to avoid these potential
shoulder pain occurring after thoracotomy in patients under adverse effects.
thoracic epidural analgesia. The high prevalence (97%) of Scawn et al3 documented that infiltration of the phrenic nerve
post-thoracotomy shoulder pain found in this study is some- at the level of the diaphragmatic periphrenic nerve fat sheath
what higher than the 42% to 86% prevalence reported by other significantly reduces the incidence of post-thoracotomy shoul-
authors in previous studies, and certainly emphasizes the im- der pain when compared with a placebo infiltration consisting
portance of the problem.2-4 Few studies have looked into dif-
ferent strategies to alleviate or decrease the incidence of post-
Table 3. Severity of Pain at Surgical Incision
thoracotomy shoulder pain.
Burgess et al2 noted an 86% prevalence of shoulder pain Group P Group A
NRS (Placebo) (Acetaminophen)
following major pulmonary resection (lobectomy, pneumonec-
tomy), while they found a low incidence (⬍7%) following 4h 2.9 ⫾ 2.9 2.0 ⫾ 2.8
minor procedures (wedge resection, biopsy). This observation 8h 1.9 ⫾ 2.3 2.3 ⫾ 2.7
led them to believe that the transection of a major airway was 12 h 1.5 ⫾ 1.9 2.0 ⫾ 2.3
16 h 1.8 ⫾ 2.3 1.6 ⫾ 2.2
the cause of this problem. Incidentally, in this study the 2
20 h 1.8 ⫾ 2.6 1.7 ⫾ 2.7
patients who did not experience shoulder pain underwent sur-
24 h 1.3 ⫾ 2.0 1.4 ⫾ 2.1
gery for a lobectomy and a pneumonectomy, respectively, and 28 h 2.1 ⫾ 2.4 1.1 ⫾ 1.7
every patient who underwent a so-called minor procedure (n ⫽ 32 h 1.6 ⫾ 2.0 1.6 ⫾ 2.3
24) experienced shoulder pain. Thus, another cause must be 36 h 1.8 ⫾ 2.2 1.2 ⫾ 1.8
considered. In the same report, Burgess et al2 suggested that 40 h 1.4 ⫾ 1.8 1.1 ⫾ 1.8
ketorolac, 30 to 60 mg intramuscularly, or ketorolac, 15 to 30 44 h 1.1 ⫾ 1.6 0.9 ⫾ 1.9
mg given intravenously, in the recovery room, followed by 48 h 0.9 ⫾ 1.4 1.0 ⫾ 1.7
ketorolac, 15 to 30 mg intramuscularly, every 6 hours for the Abbreviations: NRS, numerical rating scale, where 0 represents no
first 24 hours postoperatively, completely eliminated ipsilateral pain and 10 the worst possible pain; h, hours.
shoulder pain in 8 of 10 patients treated. No mention was made Data represent mean ⫾ SD.
478
of 0.9% saline solution. In the treated group the prevalence was
33%, compared with 85% in the control group. This study
suggests convincingly that disruption or irritation of the dia-
phragm, pericardium, or mediastinal pleural surface during
surgery may provoke pain, which is in turn referred to the
shoulder via conduction by the phrenic nerve. This positive
effect was, however, only seen in the first 2 postoperative
hours. The sympathetic nervous system could also play a role
in mediating post-thoracotomy shoulder pain. In a case report,
Garner and Coats8 showed that ipsilateral stellate ganglion
block decreased shoulder pain from a visual analog pain score
of 8 to a pain score of 3. However, since stellate ganglion block
results most of the time in a phrenic block, this would be the
most likely explanation for its proposed efficacy. A potential
problem with the block of the phrenic nerve and all approaches
resulting in a phrenic nerve block (stellate ganglion block and
interscalene brachial plexus block) is the associated diaphrag-
matic dysfunction and the resulting respiratory insufficiency.
Another suggested mechanism to explain post-thoracotomy
shoulder pain would be the pain associated with the excessive
strain of the shoulder joint capsule and ligaments that may
occur when positioning the patient in the lateral decubitus
position. However, Tan et al4 could not document the efficacy
of a suprascapular nerve block to treat ipsilateral pain occurring
after thoracotomy, discrediting the hypothesis that post-thora-
cotomy shoulder pain originates from the shoulder itself.
Marks and Brodsky,9 in a letter to the editor, suggested that
longer operations could also be associated with a higher inci-
dence of nonincisional postoperative pain among patients un-
dergoing identical operations. The results of this study suggest
that this is unlikely, because most patients had shoulder pain
and the only patients who did not have post-thoracotomy shoul-
der pain in this study underwent procedures of 100 and 125
REFERENCES
1. Salomaki TE, Laitinen JO, Nuutinen LS: A randomized double-
blind comparison of epidural versus intravenous fentanyl infusion for
analgesia after thoracotomy. Anesthesiology 75:790-795, 1991
2. Burgess FW, Anderson M, Colonna D, et al: Ipsilateral shoulder
pain following thoracic surgery. Anesthesiology 78:365-368, 1993
3. Scawn NDA, Pennefather SH, Soorae A, et al: Ipsilateral shoul-
der pain after thoracotomy with epidural analgesia: The influence of
phrenic nerve infiltration with lidocaine. Anesth Analg 93:260-264,
2001
4. Tan N, Agnew NM, Scawn ND, et al: Suprascapular nerve block
for ipsilateral shoulder pain after thoracotomy with thoracic epidural
analgesia: A double-blind comparison of 0.5% bupivacaine and 0.9%
saline. Anesth Analg 94:199-202, 2002
MAC ET AL
minutes’ duration, which is similar or even slightly above the
average duration in the whole study group.
The authors could not in the present study show a decrease
in the prevalence of shoulder pain or a difference between the
2 groups in the administration of hydromorphone to treat shoul-
der pain. This lack of difference may be attributed to the fact
that the rescue medication was used only in cases of severe
shoulder pain (NRS ⬎ 7) and that on average the patients
reported mostly a moderate intensity of pain (average NRS
3.5-4.2) . This observation regarding the severity of shoulder
pain is in accordance with that of Tan et al,4 but with somewhat
lower pain scores. In the study of Tan et al, with observations
limited to 6 hours after the performance of the suprascapular
block, the average visual analog scale (VAS) was inferior to 6,
despite high initial VAS scores (⬎7). Scawn et al3 reported
average VAS ⬍ 3.5, but the authors considered the total pain
experience and assessed it only for the first 6 hours postoper-
atively. These studies could therefore not be readily compared
with this one.
Continuous infusion of the epidural solution provided excel-
lent analgesia of the dermatomes related to the incision in the
present study, and this was confirmed by the low NRS of the
surgical incision in both groups (1.5 ⫾ 2.2 in group A and 1.7
⫾ 2.2 in group P). No difference was found in the total amount
of epidural solution administered between the 2 groups.
In conclusion, despite effective epidural analgesia, post-
thoracotomy ipsilateral shoulder pain remains a significant
problem. Preemptive and regular administration of acetamino-
phen represents a simple, secure, and inexpensive means to
reduce early (first 16 hours) postoperative ipsilateral shoulder
pain in patients undergoing thoracotomy. The natural history of
this problem reveals that the intensity of the shoulder pain
seems to decrease after the first 16 to 24 hours postoperatively.
5. Whelton A: Nephrotoxicity of nonsteroidal anti-inflammatory
drugs: physiologic foundations and clinical implications. Am J Med
106:13S-24S, 1999 (suppl)
6. Altman DG: Practical Statistics for Medical Research (ed 2).
London, UK, Chapman & Hall, 2001, pp 429-433
7. Dexter F, Chestnut DH: Analysis of statistical tests to compare
visual analog scale measurements among groups. Anesthesiology 82:
896-902, 1995
8. Garner L, Coats RR: Ipsilateral stellate ganglion block effective
for treating shoulder pain after thoracotomy. Anesth Analg 1994;78:
1195-96.
9. Marks JBD, Brodsky JB: Ipsilateral shoulder pain following
thoracic operations. II. Anesthesiology 79:192, 1993