AAAaAG

Oq7P

CSZMstas

R2F(

©P

Randomized Controlled Trial Comparingcetaminophen plus Ibuprofen Versuscetaminophen plus Codeine plus Caffeinefter Outpatient General Surgery

lex Mitchell, MD, Sander Veldhuyzen van Zanten, MD, FRCPC, Karen Inglis, RN,eoffrey Por ter, MD, FRCSC, FA C S

BACKGROUND: Narcotics are used extensively in outpatient general surger y but are often poorly tolerated withvariable efficacy. Acetaminophen combined with NSAIDs is a possible alternative. The objec-tive of this study was to compare the efficacy of acetaminophen, codeine, and caffeine ( TylenolNo. 3) with acetaminophen and ibuprofen for management of pain after outpatient generalsurger y procedures.

STUDY DESIGN: A double-blind randomized controlled trial was per formed in patients undergoing outpatientinguinal/umbilical/ventral hernia repair or laparoscopic cholecystectomy. Patients were ran-domized to receive acetaminophen plus codeine plus caffeine ( Tylenol No. 3) or acetaminophenplus ibuprofen (AcIBU) 4 times daily for 7 days or until pain-free. Pain intensity, measured fourtimes daily by visual analogue scale, was the primar y outcome. Secondar y end points includedincidence of side effects, patient satisfaction, number of days until patient was pain-free, and useof alternative analgesia.

RESULTS: One hundred forty-six patients were randomized (74 Tylenol No. 3 and 72 AcIBU), and 139 (95%)patients completed the study. No significant differences in mean or maximum daily visual analoguescale scores were identified between the 2 groups, except on postoperative day 2, when pain wasimproved in AcIBU patients (p � 0.025). During the entire week, mean visual analogue scale scorewas modestly lower in AcIBU patients (p � 0.018). More patients in the AcIBU group, comparedwith Tylenol No. 3, were satisfied with their analgesia (83% versus 64%, respectively; p � 0.02).There were more side effects with Tylenol No. 3 (57% versus 41%, p � 0.045), and the discontin-uation rate was also higher in Tylenol No. 3�treated patients (11% versus 3%, p � 0.044).

CONCLUSIONS: When compared with Tylenol No. 3, AcIBU was not an inferior analgesic and was associatedwith fe wer side effects and higher patient satisfaction. AcIBU is an effective, low-cost, and safealternative to codeine-based narcotic analgesia for outpatient general surger y procedures. ( J Am

Coll Surg 2008;206:472–479. © 2008 by the American College of Surgeons)

atsocspm

NEfCs

utpatient procedures are per formed with increasing fre-uency in general surger y; they currently constitute 60% to0% of operative procedures conducted in Nor th America.1

atients undergoing outpatient surger y require some form of

ompeting Interests Declared: None.uppor ted by Capital District Health Authority grant. Dr Por ter and Dr vananten are suppor ted by Clinical Research Scholarships of the Faculty ofedicine, Dalhousie University. MacNeil Pharmaceuticals made an unre-

tricted research grant to the Dalhousie University Depar tment of Surger y;hey were not involved in the study design, methodology, data collection, ornalysis. MacNeil Pharmaceuticals were not made aware of the results of thetudy until it was submitted for peer-revie w presentation.

eceived April 27, 2007; Revised August 30, 2007; Accepted September 5,007.rom the Departments of Surgery (Mitchell, Inglis, Porter) and Medicine

van Zanten), Dalhousie University and QEII Health Sciences Centre, Halifax, R

4722008 by the American College of Surgeons

ublished by Elsevier Inc.

nalgesia that can be taken at home in the first fe w days afterhe operation. Despite the fact that standard analgesic pre-criptions are provided to patients postoperatively, up to 40%f patients undergoing ar throscopy, gynecological laparos-opy, and carpal tunnel decompression repor ted moderate toevere pain in the first 24 hours after discharge. Paradoxically,ain medication use after outpatient surger y can be low, pri-arily as a result of side effects and fears of addiction.2

Although no population-based data examining practice

ova Scotia, Canada; Division of Gastroenterology, University of Alberta,dmonton, Alberta, Canada (van Zanten); and Department of Surgery, Hali-

ax, Nova Scotia, Canada.orrespondence address: Geoffrey Porter, MD, Department of Surgery, Dalhou-

ie University, QEII Health Sciences Center, 8-032 Centennial Bldg, 1278Tower

d, Halifax, Nova Scotia, B3H 2Y9, Canada. email: geoff.por ter@dal.ca

ISSN 1072-7515/08/$34.00doi:10.1016/j.jamcollsurg.2007.09.006

peiocmpcbdsvib

tppaemasgfbap

pcnTeNmCtft

MTdaNgP

pos

dttpbcsNclwwqw

gswissTCwtNc

t(twa

tbnfamsss

a3diis

473Vol. 206, No. 3, March 2008 Mitchell et al NSAIDS or Narcotics after Outpatient Surgery

atterns of postoperative analgesia after ambulatory gen-ral surgical procedures exist, the current standard at manynstitutions is to provide a prescription for oral acetamin-phen plus codeine (Tylenol No. 3) or oxycodone (Perco-et) for analgesia after outpatient surgery. Some patientsight receive more potent opioids, such as oral hydromor-

hone (Dilaudid). The most commonly prescribed medi-ation (Tylenol No. 3) in Canada is often poorly toleratedy patients, causing undesirable side effects. Although co-eine is a good antitussive agent, it is not very effective forevere pain.3 Adverse effects of opioids include nausea,omiting, decreased gastrointestinal motility (often result-ng in constipation), urinary retention, pruritus, increasediliary pressure, bradycardia, and altered consciousness.4

NSAIDS have been widely studied and are proved effec-ive analgesics for postoperative pain control.5,6 For exam-le, when ketorolac was compared with oral acetamino-hen and hydrocodone for tubal ligation and arthroscopy,nalgesic efficacy appeared equal, with both having a ben-fit beyond placebo.7 The hydrocodone group experiencedore side effects. Raeder and colleagues8 found ibuprofen

s effective as acetaminophen with codeine after outpatienturgery, but patients treated with ibuprofen reported betterlobal satisfaction and considerably less constipation. Un-ortunately, no randomized trials have examined the com-ination of acetaminophen plus NSAIDs compared withcetaminophen plus narcotics in outpatient general surgeryrocedures.We hypothesized that the commonly prescribed narcotic

lus acetaminophen after outpatient general surgery pro-edures might not be optimal, compared with a non-arcotic combination of an NSAID plus acetaminophen.he primary objective of this study was to compare the

fficacy of acetaminophen, codeine, and caffeine (Tylenolo. 3) with acetaminophen and ibuprofen for manage-ent of pain after outpatient general surgery procedures.omparing the two regimens in terms of side effects, pa-

ient satisfaction, number of days until patient was pain-ree, and use of alternative analgesia were secondary objec-ives of the study.

ETHODShis prospective, randomized, double-blind trial was con-ucted at a regional community hospital affiliated with ancademic tertiary care center. From February 1, 2005, toovember 30, 2005, patients aged 17 to 65 years under-

oing elective outpatient (planned discharge before 6:00M on the day of operation) unilateral inguinal hernia re-air, umbilical hernia repair, laparoscopic cholecystectomy,r small incisional hernias (� 5 cm) were eligible for the

tudy. Exclusion criteria included any preexisting pain con- d

ition requiring analgesic, fibromyalgia, recent upper gas-rointestinal bleeding, coagulopathy (primary or medica-ion related), serious renal impairment or liver disease, andregnancy. Patients with a history of peptic ulcer disease,ut currently asymptomatic, were provided with a 1-weekourse of a proton pump inhibitor to be taken during thetudy week as prophylaxis against complications fromSAIDS. Any patients with active symptomatic peptic ul-

er disease were excluded. Patients with self-described al-ergies to acetaminophen, aspirin, any NSAID, or codeineere excluded. Patients who consented to the study butho required unexpected admission, including those re-uiring admission because of operative complications,ere excluded before randomization.Eligible patients were identified by participating sur-

eons and given a brochure outlining the nature of thetudy. Patients who agreed to speak with study personnelere then contacted to establish willingness to participate

n the trial. After the initial telephone conversation withtudy personnel, additional discussion and signing of con-ent forms occurred 30 to 60 minutes before the procedure.he trial was approved by the Research Ethics Board at theapital District Health Authority (REB 2004-307) andas registered with the Food and Drug Agency and Na-

ional Institutes of Health (Clinical Trails.gov identifierCT00245375). All patients gave informed signed

onsent.All patients underwent procedures under general anes-

hetic without the use of any adjuvant regional techniqueseg, epidural, paravertebral blocks). No specific preopera-ive or intraoperative analgesic protocols were used; thisas left to the discretion of the attending surgeon and

nesthesiologist to best reflect routine surgical practice.Patients were randomized on arrival into the postanes-

hetic care unit. Stratified block randomization, using ta-les of random numbers, was used to ensure equivalentumbers of patients in each treatment group for each of theour procedures. Patients were randomized in groups of 10nd randomization was conducted using a sealed envelopeethod. The randomization code was concealed from

tudy investigators, nurses, and patients and was kept inealed envelopes in a secure location until the end of thetudy.

Patients were randomized to receive either 325 mg acet-minophen and 400 mg ibuprofen per dose (AcIBU) or00 mg acetaminophen, 15 mg caffeine, and 30 mg co-eine (the standard composition of a Tylenol No. 3 tablet

n Canada) per dose. Active medications were formulatedn powder form and then manufactured into identical cap-ules containing half of the full dose (two capsules � one

ose). Capsules were then dispensed in identical dossettes

tdgi

towspitawi“eiccfa

vutirt

mamptamrtoosdf

osAa5i

haiicrg

duatocplbmmapwawf

RFlr

(htctit(g

awarsaATt

474 Mitchell et al NSAIDS or Narcotics after Outpatient Surgery J Am Coll Surg

hat had separate compartments for each dose (28 dosesuring 7 days). In this manner, both patients and investi-ators were blinded; all packaging and capsules weredentical.

Patients were instructed (verbally and in writing) to takehe study medication (2 tablets) 4 times per day for 1 weekr until pain-free, whichever came first. Study participantsere given a diary with blank 100-mm visual analogue

cales (VAS). In this diary, patients were instructed to com-lete the VAS four times per day (at the time of each med-cation dose), to record the exact time and date at whichhey stopped taking medication (pain-free), and to recordny side effects experienced. A standard 100-mm VAS lineas presented with “no pain” at the 0 end and “worse pain

maginable” at the 100-mm end. The instructions statedplease mark on the line below how much pain you arexperiencing.” No attempts were made to focus question-ng on specific aspects of pain (eg, pain at rest, pain withough). Patients experiencing self-determined poor painontrol, or failing to tolerate medication-related side ef-ects, were able to contact study personnel and receive anlternative narcotic (hydromorphone).

Patients returned on the 7th postoperative day for a clinicisit and interview. Patients returned their diaries and un-sed medications at this time. Patients were asked abouthe incidence of specific side effects, compliance with reg-men, and other medications taken during the study pe-iod. Patients were also asked if they were satisfied withheir pain control overall (dichotomous yes/no response).

The primary end point of this study was pain intensity,easured with a 100-mm VAS (with 0 mm being no pain

nd 100 mm being maximum pain) and analyzed as dailyean and maximum scores. Some patients did not com-

lete four VAS measurements, as instructed, for each dayhey took analgesics; but in no case did a patient who tooknalgesics on a day not report at least one VAS measure-ent. In patients where four VAS measurements were not

ecorded for a given day, missing values were ignored andhe mean and maximum of daily VAS was calculated basedn the one to three measurements for that given day. Sec-ndary outcomes were incidence of side effects, patientatisfaction (yes/no), time to stopping medication, inci-ence of medication discontinuation because of side ef-ects, and need for additional analgesics.

The sample size was calculated based on the primaryutcomes of overall mean daily and maximum pain inten-ity, as measured by VAS, between the treatment arms.lthough a difference of 13 to 20 mm in VAS has beenccepted as clinically significant,9,10 we chose a threshold ofmm to ensure even a smaller efficacy difference could be

dentified statistically. A sample size of 64 in each group 2

ad 80% power to detect a difference in means of 5 mmssuming a common standard deviation of 10 mm11,12 us-ng a 2-group Student’s t-test with a 0.05 two-sided signif-cance level. Assuming losses from followup and failure toomply with study protocol of 15%, 147 patients wereequired to achieve complete data for 64 patients in eachroup.

All analyses comparing the treatment groups were con-ucted on an intention-to-treat basis. Two-tailed tests weresed at all times, and statistical significance was set a priorit p � 0.05. All univariate analyses used the Student’s-test, ANOVA (used for daily VAS data, eg, examinationf mean and maximum VAS among the four different pro-edures included in the study), or chi-square test as appro-riate. Specifically, VAS has been used extensively in the

iterature and use of the parametric Student’s t-test haseen validated for VAS pain score analysis.13,14 Overallean daily VAS and maximum daily VAS between treat-ent groups were compared using Student’s t-tests of the

ggregate respective mean and maximum daily VAS on aer-patient (not per-day) basis. Nonparametric methodsere used when the distribution of continuous data did not

pproximate the normal distribution. Statistical analysesere performed using the statistical software package SPSS

or Windows 13.0 (SPSS, Inc).

ESULTSigure 1 describes the recruitment, randomization, and fol-

owup of patients in this study. Overall, 146 patients wereandomized and formed the study cohort.

Characteristics of theTylenol No. 3 (n � 74) and AcIBUn � 72) treatment groups are shown in Table 1. Inguinalernia repair and laparoscopic cholecystectomy consti-uted the majority (80%) of procedures. Demographic andlinical variables appeared to be equally distributed be-ween the two treatment arms; specifically, no differencesn the use of local anesthetic wound infiltration, preopera-ive naproxen suppositories, operative time, or use of meshin hernia repair patients) were identified between the tworoups.

Among the 146 study patients, 140 (96%) completednd returned the study diary with VAS scores; 3 patientsere lost to followup, 2 did not complete any VAS data,

nd 1 patient withdrew from the study immediately afterandomization, without taking any study drug. Pain inten-ity, the primary outcomes measure, was reported as meannd maximum daily VAS in these 140 patients (Table 2).lthough mean daily VAS was consistently higher in theylenol No. 3 group, daily differences only reached statis-ical significance on postoperative day 2 (p � 0.025) (Fig.

). Examining the mean daily VAS during the entire 7 days

sp

fm(

eD(A

poirordm

aeisa7Neawm

tgirtc

Fsp

TC

AG

BP

OMLN

N*A

475Vol. 206, No. 3, March 2008 Mitchell et al NSAIDS or Narcotics after Outpatient Surgery

howed that patients receiving the AcIBU combination ex-erienced significantly less pain (p � 0.018).In examining the maximum daily VAS, we found similar

indings as with mean daily VAS (Table 2). On all days, theaximum daily VAS was higher in the Tylenol No. 3 group

range 1.1 mm to 7.5 mm higher), with the largest differ-

igure 1. CONSORT diagram illustrating eligible patients, exclu-ions, randomization, and followup. AcIBU, acetaminophen � ibu-rofen; T3, Tylenol No. 3 (acetaminophen � codeine � caffeine).

able 1. Clinical and Demographic Characteristics of the Stharacteristic

ge (y), mean (median; IQ range)ender, n (%)MenWomen

MI,* mean (median; IQ range)rocedure, n (%)Laparoscopic cholecystectomyInguinal herniaUmbilical herniaVentral herniaperative time (min), mean (median; IQ range)esh used, n (% of hernias)

ocal anesthetic wound infiltration, n (%)aproxen rectal suppository preoperatively, n (%)

o statistical significant differences between the 2 treatment groups (p � 0.0Calculated as kg/m2.

cIBU, acetaminophen � ibuprofen; BMI, body mass index; IQ, interquartile; Ty

nce on postoperative day 2 (7.5 mm difference; p � 0.07).uring the entire week, a nonstatistically significant trend

p � 0.08) toward lower maximum daily VAS was found incIBU patients (Fig. 3).Subgroup analyses were performed to examine the im-

act of procedure type, patient demographics, and otherperative factors on mean daily VAS. Although unilateralnguinal hernia repair and laparoscopic cholecystectomyesulted in higher mean and maximum daily VAS than thether 2 procedures, and VAS scores after ventral herniaepair were higher by the end of the study week (Fig. 4), noifferences in mean or maximum daily VAS between treat-ent arms were found in any of the four procedure groups.Secondary outcomes are summarized in Table 3. Over-

ll, 49% (n � 69) of the patients experienced at least 1 sideffect. In those randomized to receiving Tylenol No. 3, thencidence of side effects was higher (58% versus 41%, re-pectively; p � 0.04); this appeared to be mostly attribut-ble to increased nausea among these patients (16% versus%, respectively; p � 0.01). More patients in the Tylenolo. 3 arm stopped the study medication as a result of side

ffects (11% versus 3%, respectively; p � 0.04). No seriousdverse events occurred during the study; specifically, thereas no gastrointestinal bleeding, and there were only twoinor wound hematomas (one in each treatment group).Within the entire study cohort, 15 (11%) patients failed

o gain adequate pain relief within their assigned treatmentroup and required additional analgesia; 9 (13%) patientsn the Tylenol No. 3 group and 6 (9%) in the AcIBUequired additional analgesic (p � 0.43). When asked athe exit interview if they were satisfied with their painontrol (yes/no), 83% of patients receiving AcIBU stated

Cohort According to Treatment GroupAcIBU (n � 72) Tylenol No. 3 (n � 74)

48 (50; 38�56) 44 (46; 33�55)

34 (47) 33 (45)38 (53) 41 (55)29 (27; 24�31) 29 (28; 25�31)

35 (49) 35 (47)22 (31) 25 (34)12 (17) 10 (14)3 (4) 4 (5)

46 (45; 30�55) 48 (48; 32�55)29 (78) 35 (90)68 (94) 71 (96)59 (82) 61 (82)

udy

5).

lenol No. 3, acetaminophen � codeine � caffeine.

t6

sctp(t

mAt

DTawfioT

FaT

Fa

T

D

1234567O

*A ine �

476 Mitchell et al NSAIDS or Narcotics after Outpatient Surgery J Am Coll Surg

hey were satisfied with the pain control, compared with2% in the Tylenol No. 3 group (p � 0.02).

As stated previously, patients were instructed to take thetudy medications for 7 days or until pain-free, whicheverame first. Patients receiving Tylenol No. 3 stopped takinghe medication after a mean duration of 3.4 days, com-ared with AcIBU patients at a mean duration of 4.5 daysp � 0.006). This occurred despite the fact there was arend toward higher mean daily VAS on the last day of

0

5

10

15

20

25

30

35

40

45

50

0 1 2 3 4 5 6 7

Day

Mean

VA

S (

mm

)

T3AcIBU

DAY 2 p=0.025

P(overall) = 0.018

T3 71 64 51 42 33 23 18 AcIBU 69 61 55 48 43 39 32 igure 2. Pain intensity. Mean daily visual analogue scales (VAS)ccording to treatment group. AcIBU, acetaminophen � ibuprofen;

able 2. Pain Intensity as Measured by Mean and Maximum

ayn* Mean daily

AcIBU Tylenol No. 3 AcIBU Tylen

69 71 37.1 � 22.0 39.761 64 35.6 � 21.6 44.455 51 30.6 � 19.8 36.648 42 27.2 � 17.1 29.643 33 26.0 � 15.3 27.539 23 25.7 � 18.4 24.232 18 21.2 � 17.6 24.3

verall 69 71 28.8 � 16.8 36.3

Patients pain-free (no longer taking study drug) are not included.cIBU, acetaminophen � ibuprofen; Tylenol No. 3, acetaminophen � code

3, Tylenol No. 3 (acetaminophen � codeine � caffeine). T

edication use in Tylenol No. 3 patients compared withcIBU (29 mm versus 23 mm, p � 0.09; analysis restricted

o patients taking study medication fewer than 7 days).

ISCUSSIONhe goal of this study was to examine the efficacy of acet-

minophen and ibuprofen in combination when comparedith acetaminophen, codeine, and caffeine (the standard

ormulation of Tylenol No. 3). Based on the VAS painntensity data, our study demonstrated that a combinationf acetaminophen and ibuprofen is at least as effective asylenol No. 3 for pain control. Secondary outcomes, in-

0

10

20

30

40

50

60

0 1 2 3 4 5 6 7

Day

MA

X V

AS

(m

m)

T3AcIBU

Day 2 p=0.07

p (overall) = 0.08

T3 71 64 51 42 33 23 18 AcIBU 69 61 55 48 43 39 32

igure 3. Pain intensity. Maximum daily visual analogue scale (VAS)ccording to treatment group. AcIBU, acetaminophen � ibuprofen;

ily VAS (0 to 100 mm) According to Treatment GroupSD Maximum daily VAS � SD

. 3 p Value AcIBU Tylenol No. 3 p Value

.4 0.50 45.5 � 26.0 47.6 � 25.5 0.64

.6 0.025 44.8 � 23.5 52.3 � 22.7 0.07

.0 0.12 39.8 � 24.2 44.3 � 20.3 0.31

.1 0.51 33.5 � 20.8 35.6 � 19.1 0.61

.4 0.68 32.1 � 17.2 33.8 � 19.5 0.68

.1 0.76 31.0 � 21.3 31.1 � 25.0 0.76

.8 0.55 27.2 � 21.1 28.8 � 18.9 0.79

.4 0.018 36.4 � 19.6 42.5 � 20.3 0.08

caffeine; VAS, visual analogue score.

DaVAS �

ol No

� 22� 21� 19� 17� 17� 19� 16� 19

3, Tylenol No. 3 (acetaminophen � codeine � caffeine).

csf

wetVsswpT

cliymm

wtsgwdprvapwapr

pipctsttIupmwa

tiip

v

Fash

TO

A

RPTM

EA

477Vol. 206, No. 3, March 2008 Mitchell et al NSAIDS or Narcotics after Outpatient Surgery

luding incidence of side effects, patient satisfaction, andtudy medication discontinuation because of side effectsavored AcIBU.

Based on this study’s results, it is possible that a larger trialould identify more definitively statistically significant differ-

nces in pain intensity favoring AcIBU. In our opinion, such arial is unnecessary, given that a clinically relevant difference inAS is accepted to be in the range of 13 to 20 mm,9,10 and

econdary outcomes clearly favored AcIBU. Although someelect VAS measurements (eg, mean VAS postoperative day 2)ere statistically better in AcIBU patients, we prefer to em-hasize our data as not demonstrating an analgesic benefit ofylenol No. 3 over AcIBU.Why did Tylenol No. 3 patients not have better pain

ontrol? One explanation could be that the procedures se-ected were associated with relatively low levels of pain, andmproved pain control can be difficult to identify. An anal-sis of VAS for pain measurement suggested that VAS of 0m to 4 mm was equivalent to no pain, 5 mm to 44 mmild pain, 45 mm to 74 mm moderate pain, and �75 mm

0

10

20

30

40

50

1 2 3 4 5 6 7

IHR

UHR

VHR

LC

Days

VAS(meandaily)

igure 4. Pain intensity. Mean daily visual analogue scale (VAS)ccording to procedure type. IHR, inguinal hernia repair; LC, laparo-copic cholecystectomy; UHR, umbilical hernia repair; VHR, ventralernia repair.

able 3. Secondary Outcomes According to Treatment Grouutcome

ny side effect, n (%)ConstipationNauseaStomach upsetMedication stopped because of side effects, n (%)

equired other analgesics, n (%)atient satisfied with pain control, n (%)ime to stopping medications (d), mean (IQR)ean VAS (mm) on day when medication stopped (IQR)

xcludes patients lost to followup (n � 3), did not complete questionnaire (

cIBU, acetaminophen � ibuprofen; IQR, interquartile range; Tylenol No. 3, ace

as considered severe pain.15 Using this guide, most pa-ients in our trial experienced mostly mild pain and occa-ionally moderate pain. As the efficacy of AcIBU seemedreatest relative to Tylenol No. 3 on postoperative day 2,hen overall pain levels were also at their highest, our studyoes not suggest that modest increases in procedure-relatedain would result in a VAS benefit to Tylenol No. 3. Theesults seen in the Tylenol No. 3 arm are perhaps related toariable effectiveness of codeine. Codeine only works as annalgesic if metabolized by P450 enzyme CYP2D6 to mor-hine. There is a wide variation of activity of this enzymeithin a population, with 1 study demonstrating as many

s 36% of children do not metabolize codeine to mor-hine.16 In addition, when enzyme activity was quantified,educed activity was found in up to 46% of individuals.

The incidence of side effects was higher in Tylenol No. 3atients and resulted in discontinuation of the study med-cation more frequently than in the AcIBU group. Consti-ation occurred in 29% of all patients. This side effect,ommonly associated with Tylenol No. 3, was not substan-ially more common in Tylenol No. 3 patients. This mightuggest that constipation is associated with the postopera-ive course and could potentially be a result of postopera-ive physiologic changes combined with poor oral intake.nstructions and prophylaxis for constipation in patientsndergoing outpatient surgery might be indicated for allatients, not just those receiving narcotics. Nausea wasore common in patients taking Tylenol No. 3, consistentith other studies examining narcotic versus nonnarcotic

nalgesia.5,17

Only two minor wound hematomas occurred withinhis trial, one in each of the treatment groups. This makest unlikely that increased frequency of postoperative bleed-ng because of short-term use of NSAIDS is a majorroblem.Time to stopping medication within this trial data pro-

ides interesting trends. The Tylenol No. 3 group stopped

BU (n � 69) Tylenol No. 3 (n � 71) p Value

28 (41) 41 (58) 0.0417 (25) 23 (32) 0.315 (7) 11 (16) 0.014 (6) 6 (8) 0.542 (3) 8 (11) 0.046 (9) 9 (13) 0.43

57 (83) 44 (62) 0.02.5 (3�6) 3.4 (1�3) 0.00623 (7�32) 29 (13�43) 0.09

), withdrew from study (n � 1).

pAcI

4

n � 2

taminophen � codeine � caffeine; VAS, visual analogue scale.

ttpnbbpr

andta

mtscedsmqwm

soptptest8wcactTp

tlTiaT

paocptrToa[ihTqbfinnt

ocdoasbgtNs

catsusiop

A

SAA

D

478 Mitchell et al NSAIDS or Narcotics after Outpatient Surgery J Am Coll Surg

aking medication sooner, yet had higher pain scores at theime they decided to stop the medication. One could hy-othesize that the reason for stopping the medication wasot necessarily because of a lack of pain in some patientsut perhaps a reluctance to continue taking the medicationecause of its side-effect profile. This is consistent withrevious literature demonstrating side effects to be a majoreason for discontinuing pain medication.2

This study did not use any as-needed dosing of eithernalgesic regime tested. Although widely performed, as-eeded administration of analgesia appears to be substan-ard to regular dosing in randomized trials after ambula-ory surgery.18 For this reason, we elected not to includes-needed dosing within our trial design.

Patient compliance with multiple daily VAS measure-ents can be challenging. Incredibly, all patients in this

rial who took analgesics reported at least one VAS mea-urement for the given day. For the patients who did notomplete all four VAS measurements, as instructed, forach day they took analgesics, the mean and maximumaily VAS were calculated based on the one to three mea-urements for that given day. This situation represented theinority of patient evaluation days (� 20%), and its fre-

uency did not differ between treatment groups. Overall,e do not suspect a bias in efficacy evaluation related toissing VAS measurements.As mandated by the study protocol, patients stopped

tudy medication when pain-free (as depicted by the drop-ff in available patients in Figs. 2 and 3); this does limit theower of the individual VAS analyses for days later inhe week and caution should be used in interpreting thevalues for these days. Unfortunately, the standard devia-

ion found within our data (16 to 20 mm) exceeded thexpected standard deviation within our literature-basedample size planning assumptions (10 mm11,12). Even withhe higher standard deviation identified, this study has �0% power to detect overall differences of � 8 mm (stillell below the 10 to 13 mm accepted as clinically signifi-

ant9,10). Combined with the study findings, where meannd maximum daily and overall VAS favored AcIBU in allomparisons except mean VAS on day 6, we are confidenthat we have not made a type II error in concluding thatylenol No. 3 does not provide superior pain control com-ared with AcIBU.There are limitations to this study. It can be argued that

he dose of codeine (30 mg/dose) used in this study was tooow and can explain why superiority in pain control withylenol No. 3 was not identified. In fact, this trial was

nitially designed to test 60 mg codeine plus 600 mg acet-minophen plus 30 mg caffeine per dose (equivalent of 2

ylenol No. 3 tablets) compared with 400 mg ibuprofen C

lus 650 mg acetaminophen per dose. Because of a system-tic pharmacy error, the lower dosing outlined in the Meth-ds section was used in all patients. This change in medi-ations occurred inadvertently during the pharmacyreparation process and was not discovered until after therial closed and data analysis had begun; no member of theesearch team was aware of this problem during the trial.he doses patients received in this trial (325 mg acetamin-phen and 400 mg ibuprofen per dose versus 300 mg acet-minophen, 15 mg caffeine, and 30 mg codeine per doseequivalent of 1 Tylenol No. 3]) falls within the range usedn standard practice (eg, 1 or 2 Tylenol No. 3 every 6ours). In addition, if patients in our study randomized toylenol No. 3 had received 60 mg codeine per dose, it isuite possible that the incidence of side effects would haveeen more pronounced and resulted in more treatmentailures, again supporting the use of acetaminophen plusbuprofen combination. Finally, it is important to recog-ize that the regrettable systematic pharmacy error doesot impact the validity of the results found in this study forhe doses used in the two randomized arms.

This study did not use standardized preoperative, intra-perative, or postanesthesia recovery room protocols, spe-ifically around medication use. Although this was doneeliberately to mimic standard practice, it limits the abilityf the study to identify specific effective or noneffectivedjuncts to the outpatient analgesic regimens tested in thetudy. It is also important to recognize that the lack ofeneficial impact of codeine in this study might not beeneralizable to all opiates and that this study did not testhe impact of a multimodal approach of Tylenol No. 3 withSAIDS as an alternative postoperative pain control

trategy.In conclusion, acetaminophen plus ibuprofen when

ompared with Tylenol No. 3 was not an inferior analgesicnd was associated with fewer side effects and higher pa-ient satisfaction. This study failed to show any evidenceupporting the widespread use of Tylenol No. 3 in patientsndergoing outpatient general surgery procedures. The re-ults of this trial suggest that, in the current climate ofncreasing rates of narcotic misuse and abuse, alternativeutpatient analgesia, with a combination of acetamino-hen and ibuprofen, is a safe and effective strategy.

uthor Contributions

tudy conception and design: Mitchell, van Zanten, Portercquisition of data: Mitchell, Inglisnalysis and interpretation of data: Mitchell, van Zanten,

Porterrafting of manuscript: Mitchell, Porter

ritical revision: van Zanten, Inglis

AWLcftss

R

1

1

1

1

1

1

1

1

1

479Vol. 206, No. 3, March 2008 Mitchell et al NSAIDS or Narcotics after Outpatient Surgery

cknowledgment: John Murdoch, MD, FRCSC, Lesasilewski, MD, FRCSC, Warren Shih, MD, FRCSC, and

aura Nuth, MD, FRCSC contributed significantly by re-ruiting their patients to this trial. We wish to thank John Frisor data management, Hoan Linh Banh, PhD, for her assis-ance with pharmacy assistance, and the perioperative nursingtaff at the Dartmouth General Hospital for their enthusiasticupport of this study.

EFERENCES

1. Chung F, Ritchie E, Su J. Postoperative pain in ambulatorysurgery. Anesth Analg 1997;85:808–816.

2. Beauregard L, Pomp A, Choiniere M. Severity and impact ofpain after day-surgery. Can J Anaesthes 1998;45:304–311.

3. Zuckerman LA, Ferrante FM. Nonopioid and opioid analgesics.In: Ashburn MA, Rice LJ, eds. The management of pain. NewYork: Churchill Livingstone; 1998:111–140.

4. Miller RD. Anesthesia. St Louis: Mosby; 1994.5. Souter AJ, Fredman B, White PF. Controversies in the periop-

erative use of nonsterodial antiinflammatory drugs. AnesthAnalg 1994;79:1178–1190.

6. Place RJ, Coloma M, White PF, et al. Ketorolac improves recov-ery after outpatient anorectal surgery. Dis Colon Rectum 2000;43:804–808.

7. White PF, Joshi GP, Carpenter RL, et al. A comparison of oralketorolac and hydrocodone-acetaminophen for analgesia afterambulatory surgery: arthroscopy versus laparoscopic tubal liga-tion. Anesth Analg 1997;85:37–43.

8. Raeder JC, Steine S, Vatsgar TT. Oral ibuprofen versus para-

cetamol plus codeine for analgesia after ambulatory surgery.Anesth Analg 2001;92:1470–1472.

9. Kelly AM. Setting the benchmark for research in the manage-ment of acute pain in emergency departments. Emerg Med2001;13:57–60.

0. Todd KH, Funk KG, Funk JP, et al. Clinical significance ofreported changes in pain severity. Ann Emerg Med 1996;27:485–489.

1. Talbot H, Hutchinson SP, Edbrooke DL, et al. Evaluation of alocal anaesthesia regimen following mastectomy. Anaesthesia2004;59:664–667.

2. Jensen MP, Chen C, Brugger AM. Interpretation of visual ana-log scale ratings and change scores: a reanalysis of two clinicaltrials of postoperative pain. J Pain 2003;4:407–414.

3. Dexter F, Chestnut DH. Analysis of statistical tests to comparevisual analog scale measurements among groups. Anesthesiology1995;82:896–902.

4. Revill SI, Robinson JO, Rosen M, et al. The reliability of a linearanalogue for evaluating pain. Anaesthesia 1976;31:1191–1198.

5. Jensen MP, Chen C, Brugger AM. Interpretation of visual ana-log scale ratings and change scores: a reanalysis of two clinicaltrials of postoperative pain. J Pain 2003;4:407–414.

6. Williams DG, Patel A, Howard RF. Pharmacogenetics of co-deine metabolism in an urban population of children and itsimplications for analgesic reliability. Br J Anaesth 2002;89:839–845.

7. Govindarajan R, Ghosh B, Sathyamoorthy MK, et al. Efficacyof ketorolac in lieu of narcotics in the operative management oflaparoscopic surgery for morbid obesity. Surg Obes Relat Dis2005;1:530–535.

8. Pillai Riddell RR, Craig KD. Time-contingent schedules forpostoperative analgesia: a review of the literature. J Pain 2003;

4:169–175.