Oral aspirin in postoperative pain: a quantitative systematic review

Jayne E. Edwards, Anna D. Oldman, Lesley A. Smith, Dawn Carroll, Philip J. Wiffen,Henry J. McQuay, R. Andrew Moore*

Pain Research and Nuffield Department of Anaesthetics, University of Oxford, Oxford Radcliffe NHS Trust,The Churchill, Headington, Oxford OX3 7LJ, UK

Received 17 July 1998; received in revised form 22 December 1998; accepted 8 January 1999

Abstract

Objectives: A systematic review of the analgesic efficacy and adverse effects of single-dose aspirin compared with placebo in post-operative pain.Design: Published studies were identified from systematic searching of bibliographic databases and reference lists ofretrieved reports. Summed pain intensity and pain relief data were extracted and converted into dichotomous information to yield thenumber of patients with at least 50% pain relief. This was used to calculate the relative benefit and number-needed-to-treat for one patientto achieve at least 50% pain relief. For adverse effects, relative risk and number-needed-to-harm were calculated. Sensitivity analyseswere planned to test the impact of different pain models, pain measurements, sample sizes, quality of study design, and study duration onthe results.Results: Seventy-two randomized single-dose trials met our inclusion criteria, with 3253 patients given aspirin, and 3297placebo. Significant benefit of aspirin over placebo was shown for aspirin 600/650 mg, 1000 mg and 1200 mg, with numbers-needed-to-treat for at least 50% pain relief of 4.4 (4.0–4.9), 4.0 (3.2–5.4) and 2.4 (1.9–3.2) respectively. Single-dose aspirin 600/650 mg producedsignificantly more drowsiness and gastric irritation than placebo, with numbers-needed-to-harm of 28 (19–52) and 38 (22–174) respec-tively. Type of pain model, pain measurement, sample size, quality of study design, and study duration had no significant impact on theresults.Conclusions: There was a clear dose–response for pain relief with aspirin, even though these were single dose studies. Adverseeffects, drowsiness and gastric irritation were also evident in the single dose studies. The pain relief achieved with aspirin was very similarmilligram for milligram to that seen with paracetamol. 1999 International Association for the Study of Pain. Published by ElsevierScience B.V.

Keywords:Aspirin; Meta-analysis; Systematic review; Postoperative pain; Randomized controlled trial

1. Introduction

It is important for both prescribers and patients to havethe best possible information about the efficacy and safetyof analgesics. This need is reflected in patient surveys whichshow that postoperative pain is often poorly managed (Brus-ter et al., 1994). We also need to benchmark relative effi-cacy and safety of current analgesics so that we know howwell new analgesics perform. The ideal would be to judgefrom direct comparisons of the various analgesics. Unfortu-nately there are few such trials of adequate size, and hencepower, to allow credible conclusions. The alternative used

here is to determine relative efficacy indirectly, using com-parisons of different analgesics with placebo in similar clin-ical circumstances, with similar patients included, similarpain measurement and similar outcome measures, andderiving the number-needed-to-treat (Cook and Sackett,1995) for at least 50% pain relief (McQuay and Moore,1998). Using previously established methods for the con-version of pain outcome measures into dichotomous infor-mation (Moore et al., 1996, 1997ab), we have produced aquantitative systematic review of the analgesic efficacy ofaspirin, to allow comparison with other analgesics.

Meta-analyses in pain are based on a number of assump-tions such as the comparability of different acute pain mod-els, pain measurements, group sizes, the quality of studydesign, and study duration. Aspirin is widely used in clinicaltrials as a positive control, so these aspirin trials are likely to

Pain 81 (1999) 289–297

0304-3959/99/$20.00 1999 International Association for the Study of Pain. Published by Elsevier Science B.V.PII : S0304-3959(99)00022-6

* Corresponding author. Tel.: +44-1865-226-132; fax: +44-1865-226-978; e-mail: andrew.moore@pru.ox.ac.uk

be the largest data-set in acute pain. This gives the oppor-tunity to test these various assumptions. The increasedpower of meta-analysis, increased relative to the individualcomponent primary trials, also allowed us to test for factorsthat might bias interpretation of trial results (Khan et al.,1996), as well as the belief that non-steroidal anti-inflam-matory drugs have a flat dose–response curve for analgesia(Brune and Lanz, 1984).

Safety is an important factor in choosing between drugs,and aspirin in multiple-dosing has a worse gastric safetyrecord than some other non-steroidal anti-inflammatorydrugs (Henry et al., 1996). The single-dose studies in post-operative pain which are widely used for assessing analge-sic efficacy rarely generate sufficient safety data, and someadverse effects may only be detected in chronic use. Thelarge set of single-dose aspirin studies allowed us to explorethe idea that adverse effects, such as gastric irritation, maybe evident in single-dose use, but only when the results frommany trials are combined.

2. Methods

Single dose, randomized controlled trials of aspirin inpostoperative pain (post dental extraction, postsurgical orpostpartum pain) were sought. Different search strategieswere used to identify eligible reports from MEDLINE(1966–March 1998), EMBASE (1980–Jan 1998), theCochrane Library (Issue 1, 1998), and the Oxford PainRelief Database (1950–1994) (Jadad et al., 1996a). A freetext search was undertaken using the terms ‘aspirin’, ‘acetylsalicylic acid’, ‘clinical trial’, ‘trial’, ‘study’, ‘random*’,‘double blind’, ‘analgesi*’, ‘pain*’, ‘post-operat*’,‘surg*’, ‘dental’, ‘molar’, ‘extraction’, ‘operat*’, ‘postsur-gical’. Variants of the individual terms were also used, forexample ‘post-surgical’ and ‘postsurgical’. No restrictionsto language were made. Additional reports were identifiedfrom the reference lists of retrieved reports. Neither phar-maceutical companies nor authors of papers were contactedfor unpublished reports. Abstracts and review articles werenot sought. The inclusion criteria used were: randomizedallocation to treatment groups which included aspirin andplacebo, full journal publication, postoperative oral admin-istration, adult patients, baseline pain of moderate to severeintensity (which for visual analogue scales (VAS) equates to.30 mm (Collins et al., 1997)), double blind design and anestablished measure of postoperative pain. Acceptable painmeasurements were: (i) a five-point pain relief scale withstandard wording (none, slight, moderate, good, complete);or (ii) a four-point pain intensity scale (none, mild, moder-ate or severe); or (iii) a VAS for pain relief or pain intensity;or iv) total pain relief (TOTPAR) or summed pain intensitydifference (SPID) or their visual analogue equivalents(VASTOTPAR or VASSPID) at 4, 5, or 6 h (or sufficientdata provided to allow their calculation).

We excluded reports of aspirin for the relief of other pain

conditions, aspirin in combination with other drugs, trialswhich reported data from a cross-over design as a singledata set, trials where the number of patients per treatmentgroup was less than ten (L’Abbe et al., 1987), and trialswhich included pain relief data collected after additionalanalgesic was given. Trials using enteric-coated, sustainedrelease or suspension formulations were excluded; tabletformulation was assumed when formulation was not stated(Table 1).

Each report which could possibly be described as a ran-domized controlled trial was read independently by at leastthree authors (JE, AO, LS) and scored using a commonly-used three item, 1–5 score, quality scale (Jadad et al.,1996b). Consensus was then achieved. The maximumscore of an included study was five and the minimumscore was two.

2.1. Statistical analyses

From each report we took: the numbers of patients trea-ted, the mean TOTPAR, SPID, VASTOTPAR or VAS-SPID, study duration and the dose given. Information onadverse effects was also extracted. For each report, themean TOTPAR, SPID, VASTOTPAR or VASSPID values

Table 1

Study characteristics

Number of comparisons Number of patients

Dose300/325 2 156500 3 250600/650 68 5069900 2 801000 8 7161200 5 279

Number of comparisons %

Pain modelDental 60 68Episiotomy 11 13Post surgical 12 14Mixed 5 5FormulationTablet/capsule 58 66Buffered 7 8Soluble 4 4Did not state 19 22Duration (h)4 21 245 18 206 49 56Quality score5 20 234 30 343 35 402 3 3LanguageEnglish 86 98Other 2 2

290 J.E. Edwards et al. / Pain 81 (1999) 289–297

for active and placebo were converted to the percentage ofthe maximum possible TOTPAR or SPID, %maxTOTPARor %maxSPID, by division by the calculated maximumvalue (Cooper, 1991). The proportion of patients in eachtreatment group who achieved at least 50% of the maximumTOTPAR was calculated using verified equations (Moore etal., 1996; Moore et al., 1997a; Moore et al., 1997b). Theseproportions were then converted into the number of patientsachieving at least 50% of the maximum TOTPAR by multi-plying by the total number of patients in the treatmentgroup. Information on the number of patients with at least50% of the maximum TOTPAR for active and placebo wasthen used to calculate relative benefit (RB) and number-needed-to-treat (NNT).

Relative benefit and relative risk (RR) estimates werecalculated with 95% confidence intervals (CI) using afixed effects model (Gardner and Altman, 1986). NNT ornumber-needed-to-harm (NNH) with 95% confidence inter-vals were calculated by the method of Cook and Sackett(1995) when the relative benefit or relative risk estimateswere statistically significant. The confidence interval of theNNT indicates no benefit of one treatment over the otherwhen the upper limit includes infinity. A statistically sig-nificant difference from control was assumed when the 95%confidence interval of the relative benefit did not includeone.

Calculations were performed using Excel v. 5.0 and Stat-View v. 4.5 on a Power Macintosh 8500/150.

Efficacy analysis was carried out for all dose groups withsufficient sample size (a minimum of three trials). Sensitiv-ity analyses were performed using the largest data set(aspirin 600 mg combined with aspirin 650 mg). Graphicalanalysis and analysis of variance were used for the sensitiv-ity analyses.

3. Results

One hundred and seventy-five publications were identi-fied. Of these, six could not be obtained from the BritishLibrary (Gallardo et al., 1982, 1984; Cordero, 1985;Mandujano; Or and Bozkurt, 1985; Carstens et al., 1987).Of the remaining studies 31 were not placebo controlled, 16

failed to meet our inclusion criteria for baseline pain, 16 hadno extractable data, 13 used an inappropriate study design,eight were not randomized, seven used both postoperativeand trauma pain, five were duplicate publications (DeVroey, 1978; Honig, 1978; Rye and Siegel, 1978; Cooper,1980; Cooper et al., 1986;), two used excluded formulationsof aspirin, and two were not double-blind. The references tothese excluded trials are available at http://www.jr2.ox.a-c.uk/Bandolier/painres/aspac/aspac.html.

Sixty-nine publications, reporting on 72 trials, met ourinclusion criteria. These trials generated a total of 88 aspirinversus placebo comparisons, providing data on 6550patients (3253 received aspirin, and 3297 placebo). Themedian quality score for the trials was four. Dental surgerywas the setting for 68% of the comparisons, 66% stated thatthey used tablets or capsules, 56% were 6-h studies and 98%were published in English. Further study characteristics arelisted in Table 1, and full details of the included trials areavailable at http://www.jr2.ox.ac.uk/ Bandolier/painres/aspac/aspac.html.

3.1. Efficacy analysis

Insufficient data were available for analysis of aspirin300/325 mg and 900 mg.

3.1.1. Aspirin 500 mg (256 patients)The proportion of patients with at least 50% pain

relief varied between 26% and 46% (mean 34%) for as-pirin 500 mg, and 20–32% (mean 28%) for placebo.The relative benefit of aspirin 500 mg versus placebowas 1.2 (0.8–1.8), indicating no statistically significant ben-efit of this dose of aspirin over placebo in these trials (Table2).

3.1.2. Aspirin 600/650 mg (5069 patients)Fig. 1 (top panel) shows the proportion of patients with at

least 50% pain relief on aspirin 600/650 mg (11–90%, mean40%) compared with placebo (0–50%, mean 17%), relativebenefit 2.0 (1.8–2.2). For a single dose of aspirin 600/650mg compared with placebo, the number-needed-to-treat forat least 50% pain relief was 4.4 (4.0–4.9) over 4–6 h (Table2).

Table 2

Summary of relative benefit and number-needed-to-treat for trials of aspirin compared with placebo

Aspirin dose (mg) Number ofcomparisons

Patients with atleast 50% painrelief with aspirin

Patients with atleast 50% painrelief with placebo

Relativebenefit (98% CI)

NNT (95% CI)

500 3 45/135 32/115 1.2 (0.8–1.8) nc600/650 68 960/2499 404/2562 2.0 (1.8–2.2) 4.4 (4.0–4.9)1000 5 153/357 64/359 2.2 (1.4–3.4) 4.0 (3.2–5.4)1200 5 85/140 27/139 3.3 (1.8–6.3) 2.4 (1.9–3.2)

nc, Not calculated because relative risk not statistically significant.

291J.E. Edwards et al. / Pain 81 (1999) 289–297

3.1.3. Aspirin 1000 mg (716 patients)Fig. 1 (middle panel) shows the proportion of patients

with at least 50% pain relief on aspirin 1000 mg (24–65%, mean 44%) compared with placebo (0–32%, mean20%), relative benefit 2.2 (1.4–3.4). For a single dose ofaspirin 1000 mg, compared with placebo, the number-needed-to-treat for a least 50% pain relief was 4.0 (3.2–5.4) over 4–6 h (Table 2).

3.1.4. Aspirin 1200 mg (279 patients)Fig. 1 (bottom panel) shows the proportion of patients

with at least 50% pain relief on aspirin 1200 mg (50–75%, mean 62%) compared with placebo (7–36%, mean19%), relative benefit 3.3 (1.8–6.3). For a single dose ofaspirin 1200 mg compared with placebo, the number-needed-to-treat for a least 50% pain relief was 2.4 (1.9–3.2) over 4–6 h (Table 2).

The efficacy dose–response for aspirin is shown in Fig. 2

together with the dose–response for paracetamol (Moore etal., 1997c).

3.2. Sensitivity analyses

Trials of aspirin 600/650 mg versus placebo were used forthe sensitivity analyses. Numbers-needed-to-treat foranalgesic efficacy, at least 50% pain relief single-dose post-operative pain for 4–6 h, compared with placebo, werecalculated by group size, quality score, pain model, typeof pain measurement, and study duration.

3.2.1. Group sizeThe median aspirin group size for all comparisons was 38

patients. For group sizes below the median the number-needed-to-treat for at least 50% pain relief was 4.1 (3.5–5.0) compared with 4.6 (4.1–5.3) for group sizes greaterthan or equal to 38. There was no significant differencebetween the numbers-needed-to-treat of the larger and smal-ler group sizes (P = 0.8).

3.2.2. Quality scoreTrials with quality scores of three and above were com-

Fig. 1. L’Abbe plots of the results from the aspirin trials included in thereview. Percentage of patients achieving at least 50% pain relief on aspirin

Fig. 2. Dose–response curves for aspirin and paracetamol (paracetamoldata from (Moore et al., 1997c)). Numbers indicate the number of studiesfor each dose.

292 J.E. Edwards et al. / Pain 81 (1999) 289–297

pared with those with a quality score of two. Little differ-ence was evident between the NNTs for studies with differ-ent quality scores (Fig. 3).

3.2.3. Pain modelTrials in dental pain were compared with all the other

pain models (episiotomy, post-surgical and trauma pluspost-surgical models) (Fig. 3). There was no statisticallysignificant difference between the number-needed-to-treatof 4.7 in dental pain (95%CI, 4.2–5.4) and the number-needed-to-treat of 3.9 for the other pain models (3.3–4.7)(Fig. 3).

3.2.4. Pain measurement and study durationThere was no significant difference between the numbers-

needed-to-treat for the different pain measurements or forthe different study durations (Fig. 4).

3.3. Adverse effects

3.3.1. Total adverse effectsSixty of the 72 trials reported adverse effect information

for the various doses of aspirin and placebo. Six trialsreported no adverse effects with either aspirin or placebo.A total of 313/2619 (12%) of the patients on aspirin (alldoses) and 261/2660 (10%) of the patients on placeboreported at least one adverse effect (Table 3), relative risk1.3 (0.9–1.5). The most frequently reported effects weredizziness, drowsiness, gastric irritation, nausea and vomit-ing. All effects were of mild or moderate severity and nopatients withdrew as a result.

3.3.2. Analysis by dose of aspirinAmong the various doses tested in the trials, there was

sufficient information for the analysis of individual adverseeffects only for aspirin 600/650 mg compared with placebo.A total of 257/1976 (13%) of the patients on aspirin 600/650mg and 229/2088 (11%) of the patients on placebo reportedan adverse effect, relative risk 1.2 (1.03–1.4). The number-needed-to-harm was 44 (23–345) for aspirin 600/650 mgcompared with placebo.

Significantly higher incidences of drowsiness andgastric irritation were reported with aspirin 600/650 mgthan with placebo, numbers-needed-to-harm 28 (19–52) and 38 (22–174) respectively (Table 3). No significantdifference between aspirin 600/650 mg and placebo wasshown for nausea, vomiting, dizziness or headache (Table3).

4. Discussion

Aspirin has been known to be an effective analgesic formany years. Rheumatism has affected man since the greatriver cultures of the Middle East. Clay tablets from theSumerian period described the use of willow leaves to

treat it. The Egyptians were also aware of the pain-relievingeffects of potions made from myrtle or willow leaves.Edward Stone, a vicar from Chipping Norton in Oxford-shire, is generally recognized as the man who gave thefirst scientific description of the effects of willow bark. In1763 he wrote a letter to the Earl of Macclesfield, thenPresident of the Royal Society in London, in which hedescribed treating patients suffering from ague (fever)with 20 grains (approximately a gram) of powdered willowbark in a dram of water every 4 h.

The results of the present systematic review confirmaspirin’s efficacy as a single dose postoperative analgesic.The analgesic dose–response tells us that bigger doses givemore analgesia, and comparing aspirin with paracetamol theanalgesia produced by the two drugs is very similar. Onegram of aspirin gives the same analgesia as one gram ofparacetamol (Fig. 2). This mg for mg equivalence is thesame conclusion reached by Houde and Wallenstein aftertheir randomized comparison of 600 mg of aspirin and 600mg paracetamol, cited in (Houde et al., 1965).

Houde and Wallenstein also quite correctly emphasizedthat with NSAIDs (unlike opioids) it was often very difficultto show an analgesic dose–response for outcomes such aspeak relief. When differences were apparent in single trials,they were seen using total pain relief. This meta-analysisshows that there is indeed an underlying dose–response, andwe used total pain relief (area-under-the-curve for painrelief against time). As Fig. 2 shows, the slope of the

Fig. 3. Sensitivity analyses: number-needed-to-treat (NNT) by qualityscore (top panel) and by pain model (bottom panel).

293J.E. Edwards et al. / Pain 81 (1999) 289–297

dose–response is very similar for aspirin and paracetamol.The area-under-the-curve for pain relief against time mayconceal differences between formulations of a particulardrug, and indeed difference between drugs. Fast-onsetfast-offset medications may theoretically underperformcompared with slower-onset longer-offset alternatives.That question could not be addressed adequately from thisdata. Whether it is clinically relevant is also unclear. Withinthe aspirin trials we saw little indication of substantial dif-ferences in time-effect curves between formulation, orbetween drug. Extreme differences in time-effect curveswould clearly require us to investigate with a finer toolthan the area-under-the-curve, for pain relief against time.For the present we believe that the NNT derived from thatarea is a robust and clinically useful comparator.

The importance of this aspirin dose–response, and thatfound with similar analytic methods for ibuprofen but notdiclofenac (Collins et al., 1998), is that it suggests thatpotentially safer NSAIDs, such as COX-2 drugs, mightallow us to exploit the analgesic potential of higherNSAID doses. We do not know whether higher doses ofNSAIDs alone could replicate the analgesia which highdoses of opioid alone can achieve. Injecting 20 mg of mor-phine has a NNT of less than two, better than the NNT of 10mg of morphine (2.9; 2.6–3.6) (McQuay et al., 1999; Nor-holt et al., 1996). If higher doses of COX-2 NSAIDs canemulate this feat, this opens up the prospect of NSAIDreplacing opioid for certain indications.

A second intriguing feature of this review is the informa-tion about adverse effects. Single-dose trials of analgesicsare usually too small to pick up rare, but severe adverseeffects like the hepatic problems with bromfenac (FDATalk Paper, 1998). They are also usually too small to giveus credible incidence data for more common adverseeffects. A third problem is that we are sceptical of the pre-dictive value of single dose trials for the adverse effects thatbecome apparent with chronic use. Within this meta-analy-sis, however, is an indication that pooled data from manysingle-dose trials may be a better predictor than the compo-nent trials alone. Gastric irritation showed up as a significantadverse effect so that about one patient in 40 will havegastric irritation from a single dose of aspirin (Table 3).

Another view of the same problem is that about one elderlyperson will be admitted to hospital with ulcer bleeding per3000 NSAID prescriptions (Hawkey et al., 1997). We do notknow the predictive value of single dose gastric irritation forthe chronic problem, but we do know that we can detectsingle dose gastric irritation by pooling data from multiplesmall trials. The significant drowsiness seen with aspirin600/650 mg, NNH 28 (19–52), has been seen with a similaranalysis for ibuprofen (Edwards et al., 1999), with higherincidence after dental surgery than after other procedures.

Imagine a meta-analysis of an analgesic, in which wepooled data from different postoperative pain states, suchas dental or orthopaedic surgery, or pooled data from trialswhich used different pain measurements. Any conclusions

Table 3

Adverse effects: aspirin 600/650 mg compared with placebo

Dose Numberof trials

Patients with adverseeffects with aspirin

Patients with adverseeffects with placebo

Relative risk(95% CI)

NNH (95% CI)

All doses Total adverse effects 60 313/2619 261/2660 1.3 (0.0–1.5) NCAspirin 600/650 mg Total adverse effects 53 257/1976 229/2088 1.2 (1.03–1.4) 44 (23–345)

Dizziness 30 41/1429 27/1557 1.6 (0.9–2.6) ncDrowsiness 33 103/1542 56/1672 1.9 (1.4–2.5) 28 (19–52)Gastric irritation 11 20/546 6/562 2.5 (1.2–5.1) 38 (22–174)Headache 29 34/1237 56/1363 0.7 (0.4–1.02) ncNausea 34 54/1563 68/1683 0.8 (0.6–1.2) ncVomiting 21 12/835 18/927 0.7 (0.4–1.6) nc

nc, Not calculated because relative risk not statistically significant.

Fig. 4. Sensitivity analyses: number-needed-to-treat (NNT) by pain out-come (top panel) and by duration of study observations (bottom panel).

294 J.E. Edwards et al. / Pain 81 (1999) 289–297

might be said to be faulty because apples were pooled withoranges, rather than pooling either apples alone or orangesalone. This review provides empirical evidence from aspirinthat results will be the same for different acute pain states,different pain measurements, different group sizes, differentquality of study design, and different study durations. Thisshould help future reviewers who wish to study interven-tions in postoperative pain. It is important empirical supportfor the ‘lumpers’, who wish pragmatically to pool all admis-sible data, against a ‘splitter’ contention that pain is differ-ent at different sites and responds differently to analgesics.Common sense should operate. This is not a proposal toreplace injected opioid on the day of major surgery withoral aspirin. Rather the evidence suggests that trials thattested oral aspirin in different clinical settings came upwith very similar efficacy, despite the difference in clinicalsetting. A second cautionary note is that although thisreview identified a very large number of papers and hencepatients compared with similar reviews of other analgesics,we had to discard two-thirds of the papers we found becausethey did not meet the inclusion criteria. The confidenceintervals around answers to important questions of efficacyand safety remain wide, despite a unique wealth of informa-tion.

Acknowledgements

This study was supported by grants from NHS R and DHealth Technology Evaluation programmes (#93/31/4 and#94/11/4), European Union Biomed 2 BMH4 CT95 0172,SmithKline Beecham Consumer Healthcare, the Biotech-nology and Biological Sciences Research Council, andPain Research Funds. David Gavaghan and Martin Tramerprovided very helpful comments.

Appendix A. Included reports

[1] Beaver, W.T., Forbes, J.A. and Shackleford, R.W., A method for the12-hour evaluation of analgesic efficacy in outpatients with postoperativeoral surgery pain, Three studies of diflunisal, Pharmacotherapy, 3 (1983)23S–37S.

[2] Bloomfield, S.S., Gaffney, T.E. and Howet, M., Comparativeanalgesic efficacy of chlorphenesin carbamate and acetylsalicylic acidafter episiotomy, Anesth. Analg., 46 (1967) 515–520.

[3] Boraks, S., Flurbiprofen in low dosage compared with metamizolesodium (dipyrone), acetylsalicylic acid and placebo in treatment of painfollowing dental extractions, Arq. Bras. Med., 61 (1987) 424–430.

[4] Breivik, H., Stenseth, R., Apalseth, K. and Spilsberg, A.M., Pirox-icam, acetylsalicylic acid and placebo for postoperative pain, Acta Anaes-thesiol. Scand., 28 (1984) 37–39.

[5] Calimlim, J.F., Wardell, W.M., Davis, H.T., Lasagna, L. and Gillies,A.J., Analgesic efficacy of an orally administered combination of nenta-zocine and aspirin-with observations on the use and statistical efficiency ofglobal subjective efficacy ratings, Clin. Pharmacol. Ther., 21 (1977) 34–43.

[6] Clark, M.S., Lindenmuth, J.E., Silverstone, L.M. and Fryer, G.E.Jr.,A double blind single dose evaluation of the relative analgesic efficacy andsafety of carprofen in the treatment of postoperative pain after oral surgery,Oral Surg. Oral Med. Oral Pathol., 68 (1989) 273–278.

[7] Cooper, S.A., Breen, J.F. and Giuliani, R.L., Replicate studiescomparing the relative efficacies of aspirin and indoprofen in oral surgeryoutpatients, J. Clin. Pharmacol., 19 (1979) 151–159.

[8] Cooper, S.A., Engel, J., Ladov, M., Precheur, H., Rosenheck, A. andRauch, D., Analgesic efficacy of an ibuprofen-codeine combination, Phar-macotherapy, 2 (1982) 162–167.

[9] Cooper, S.A., Hutton, C., Reynolds, D.C., Gallegos, L.T., Allen, C.,Marriott, J.G. and Ciccone, P.E., Dose–response analgesic activity ofmeclofenamate sodium in dental pain, Adv. Ther., 8 (1991) 157–165

[10] Cooper, S.A., Itkin, A. and Zweig, B., Comparison of oxaprozin,aspirin, and placebo in a dental impaction pain model, Adv. Ther., 9(1992) 184–194.

[11] Cooper, S.A. and Mardirossian, G., Comparison of flurbiprofenand aspirin in the relief of postsurgical pain using the dental pain model,Am. J. Med., 80 (1986) 36–40.

[12] Cooper, S.A., Mardirossian, G. and Milles, M., Analgesic relativepotency assay comparing flurbiprofen 50, 100, and 150 mg, aspirin 600mg, and placebo in postsurgical dental pain, Clin. J. Pain, 4 (1988) 175–81.

[13] Cooper, S.A., Needle, S.E. and Kruger, G.O., Comparative anal-gesic potency of aspirin and ibuprofen, J. Oral Surg., 35 (1977) 898–903.

[14] Cooper, S.A., Wagenberg, B., Eskow, R. and Zissu, J., Doubleblind evaluation of suprofen and aspirin in the treatment of periodontalpain, Pharmacology, 27 Supp. (1983) 23–30.

[15] Cooper, S.A., Wagenberg, B., Zissu, J., Kruger, G.O., Reynolds,D.C., Gallegos L.T., Friedmann, N. and Darnna, R.P., The analgesic effi-cacy of suprofen in peridontal and oral surgical pain, Pharmacotherapy, 6(1986) 267–276.

[16] Coutinho, A., Bonelli, J. and Nanci de Carvelho, P., A double-blind study of the analgesic effects of fenbuen, codeine, aspirin, propox-yphene and placebo, Curr. Ther. Res., 19 (1976) 58–65.

[17] Desjardins, P.J., Cooper, S.A., Gallego, T.L., Allwein, J.B., Rey-nolds, D.C., Kruger, G.O. and Beaver, W.T., The relative analgesic effi-cacy of propiram fumarate, codeine, aspirin, and placebo in post impactiondental pain, J. Clin. Pharmacol., 24 (1984) 35–42.

[18] De Vroey, M.D., Steelman, S.L., Caudon, J., Verhaest, L., Besse-laar, G.H. and Buntinx, A., A double-blind, placebo-controlled, single-dose study comparing three-dose levels of diflunisal with aspirin and pla-cebo in patients with pain due to episiotomy, Acta Therapeutica, 3 (1977)205–215.

[19] Fliedner, L., Levsky, M. and Kechejian, H., Analgesia with eto-dolac in oral postsurgical pain, Curr. Ther. Res., 36 (1984) 33–45.

[20] Forbes, J.A., Barkaszi, B.A., Ragland, R.N. and Hankle, J.J.,Analgesic effect of fendosal, ibuprofen and aspirin in postoperative oralsurgery pain, Pharmacotherapy, 4 (1984) 385–391.

[21] Forbes, J.A., Beaver, W.T., Jones, K.F., Edquist, I.A., Gongloff,C.M., Smith, W.K., Smith, F.G. and Schwartz, M.K., Analgesic efficacy ofbromfenac, ibuprofen, and aspirin in postoperative oral surgery pain., Clin.Pharmacol. Ther., 51 (1992 Mar) 343–352.

[22] Forbes, J.A., Butterworth, G.A., Burclfield, W.H. and Beaver,W.T., Evaluation of ketorolac, aspirin, and an acetaminophen-codeinecombination in postoperative oral surgery pain, Pharmacotherapy, 10(1990) 77S–93S.

[23] Forbes, J.A., Butterworth, G.A., Burclfield, W.H., Beaver, W.T.and Shackleford, R.W., A 12 hour evaluation of the analgesic efficacy ofdiflunisal, zomepirac sodium, aspirin, and placebo in postoperative oralsurgery pain, Pharmacotherapy, 3 (1983) 38S–46S.

[24] Forbes, J.A., Calderazzo, J.P., Bowser, M.W., Foor, V.M., Shack-leford, R.W. and Beaver, W.T., A 12 hour evaluation of the analgesic

295J.E. Edwards et al. / Pain 81 (1999) 289–297

efficacy of diflunisal, aspirin, and placebo in postoperative dental pain, J.Clin. Pharmacol., 22 (1982) 89–96.

[25] Forbes, J.A., Edquist, I.A., Smith, F.G., Schwartz, M.K. and Bea-ver, W.T., Evaluation of bromfenac, aspirin, and ibuprofen in postopera-tive oral surgery pain, Pharmacotherapy, 11 (1991) 64–70.

[26] Forbes, J.A., Jones, K.F., Kehm, C.J., Smith, W.K., Gongloff,C.M., Zeleznock, J.R., Smith, J.W., Beaver, W.T. and Kroesen, M., Eva-luation of aspirin, caffeine, and their combination in postoperative oralsurgery pain, Pharmacotherapy, 10 (1990) 387–93

[27] Forbes, J.A., Keller, C.K., Smith, J.W., Zeleznock, J.R., Sevelius,H. and Beaver, W.T., Analgesic effect of naproxen sodium, codeine, anaproxen codeine combination and aspirin on the postoperative pain oforal surgery, Pharmacotherapy, 6 (1986) 211–218.

[28] Forbes, J.A., White, R.W., White, E.H. and Hughes, M.K., Anevaluation of the analgesic efficacy of proquazone and aspirin in post-operative dental pain, J. Clin. Pharmacol., 20 (1980) 465–474.

[29] Forbes, J.A., Yorio, C.C., Selinger, L.R., Rosenmertz, S.K. andBeaver, W.T., An evaluation of flurbiprofen, aspirin, and placebo in post-operative oral surgery pain, Pharmacotherapy, 9 (1989) 66–73.

[30] Frame, J.W., Fisher, S.E., Pickvance, N.J. and Skene, A.M., Adouble blind placebo controlled comparison of three ibuprofen combina-tions and aspirin, Br. J. Oral Maxo. Surg., 24 (1986) 122–129.

[31] Gaston, G.W., Mallow, R.D. and Frank, J.E., Comparison of eto-dolac, aspirin and placebo for pain after oral surgery, Pharmacotherapy, 6(1986) 199–205.

[32] Hebertson, R.M. and Storey, N., The comparative efficacy ofdiclofenac potassium, aspirin, and placebo in the treatment of patientswith pain following gynecologic surgery, Today’s Ther. Trends, 12(1994) 33–45.

[33] Herrmann, U., Laube, W., Roth, F., Schultheiss, H.R., Berger, M.,Schurmann, P., Von Graffenried, B. and Nuesch, E., Analgesic effect offluproquazone in postoperative patients, Clin. Pharmacol. Ther., 27 (1980)379–385.

[34] Holland, I.S., Seymour, R.A., Ward Booth, R.P., Ord, R.A., Lim,K.L. and Hoare, R.C., An evaluation of different doses of soluble aspirinand aspirin tablets in postoperative dental pain, Br. J. Clin. Pharmacol., 26(1988) 463–468.

[35] Honig, W.J., Cremer, C.W. and Manni, J.G., A single dose studycomparing the analgesic effects of diflunisal, acetylsalicylic acid, andplacebo in pain following meniscectomy, J. Int. Med. Res., 6 (1978)172–179.

[36] Jain, A.K., Hunley, C.C., Kuebel, J., McMahon, F.G. and Ryan,J.J., Analgesic efficacy of amfenac, aspirin and placebo after extraction ofimpacted teeth, Pharmacotherapy, 6 (1986) 236–240.

[37] Jain, A. K., McMahon, F.G. and Ryan, J.R., An analgesic study ofindoprofen, aspirin and placebo in dental surgery patients, Curr. Ther.Res., 38 (1985) 682–688.

[38] Jain, A.K., McMahon, F.G., Ryan, J.R. and Smith, G.B., Analgesicefficacy of indoprofen in postpartum episiotomy pain, Curr. Ther. Res.Clin. Exp., 38 (1985) 677–681.

[39] Jain, A.K., Ryan, J.R., McMahon, F.G., Kuebel, J.O., Walters, P.J.and Noveck, C., Analgesic efficacy of low dose ibuprofen in dental extrac-tion pain, Pharmacotherapy, 6 (1986) 318–322.

[40] Kantor, T.G., Sunshine, A., Laska, E., Meisner, M. and Hopper,M., Oral Analgesic Studies: pentazocine hydrochloride, codeine, aspirin,and placebo and their influence on response to placebo, Clin. Pharmacol.Ther., 7 (1965?) 447–454.

[41] Kempf, K.K., Konzelman, J.L., Schultz, R.E. and Turner, J.L.,Comparison of meclofenamate sodium with buffered aspirin and placebofor the relief of postoperative dental pain, Clin. Ther., 9 (1987) 594–601

[42] Lehnert, S., Reuther, J., Wahl, G., Brarthel, K., Wirksamkeit vonparacetamol (tylenol) und acetylsalizylsaure (aspirin) bei postperativenschmerzen, Dtsch Zahnarztl Z, 45 (1990) 23–26.

[43] Lindenmuth, J.E., Clark, M.S. and Fryer, G.E. Jr,, Evaluation of

the analgesic efficacy and safety of carprofen, Anesth. Prog., 36 (1989)206–209.

[44] London, R., Sundaram, G.S., Feldman, S. and Goldstein, P., Epi-siotomy pain efficacy and safety of fluproquazone compared to aspirin andplacebo, Int. J. Gynaecol. Obstet., 21 (1983) 251–255.

[45] London, R.S., Sundaram, G.S., Feldman, S. and Goldstein, P.J.,Aspirin in the treatment of episiotomy pain, South Med. J., 76 (1983) 844–845.

[46] Mardirossian, G. and Cooper, S.A., Comparison of the analgesicefficacy of flurbiprofen and aspirin for postsurgical dental pain, J. OralMaxillofac. Surg., 43 (1985) 106–9.

[47] Markowitz, N.R., Young, S.K., Rohrer, M.D. and Turner, J.L.,Comparison of meclofenamate sodium with buffered aspirin and placeboin the treatment of postsurgical dental pain, J. Oral. Maxillofac. Surg., 43(1985) 517–522.

[48] McQuay, H.J., Carroll, D., Poppleton, P., Summerfield, R.J. andMoore, R.A., Fluradoline and aspirin for orthopedic postoperative pain,Clin. Pharmacol. Ther., 41 (1987) 531–536.

[49] Mehlisch, D.R. and Brown, P., Single-dose therapy with diclofe-nac potassium aspirin, or placebo following dental impaction surgery,Today’s Ther. Trends, 12 (1994) 15–31.

[50] Mehlisch, D.R. and Frakes, L.A., A controlled comparative eva-luation of acetaminophen and aspirin in the treatment of postoperativepain, Clin. Ther., 7 (1984) 89–97.

[51] Mehlisch, D.R., Sterling, W.R., Mazza, F.A. and Singer, J.M., Asingle dose study of the efficacy and safety of FS 205 397 (250 mg or 500mg) versus aspirin and placebo in the treatment of postsurgery dental pain,J. Clin. Pharmacol., 30 (1990) 815–823.

[52] Nelson, S. and Brahim, J., An evaluation of the analgesic efficacyof diclofenac potassium, aspirin, and placebo in postoperative dental pain,Today’s Ther. Trends, 12 (1994) 3–14.

[53] Nelson, S.L. and Bergman, S.A., Relief of dental surgery pain: acontrolled 12 hour comparison of etodolac, aspirin, and placebo, Anesth.Prog., 32 (1985) 151–156.

[54] Nelson, S.L., Brahim, J.S., Korn, S.H., Greene, S.S. and Suchower,L.J., Comparison of single-dose ibuprofen lysine, acetylsalicylic acid, andplacebo for moderate-to-severe postoperative dental pain, Clin. Ther., 16(1994) 458–465.

[55] Okun, R., Green, J.W. and Shackleford, R.W., An analgesic com-parison study of indoprofen versus aspirin and placebo in surgical pain, J.Clin. Pharmacol., 19 (1979) 487–492.

[56] Or, S. and Bozkurt, A., Analgesic effect of aspirin, mefenamic acidand their combination in post operative oral surgery pain, J. Int. Med. Res.,16 (1988) 167–172.

[57] Parkhouse, J., Rees-Lewis M., Skolinik, M. and Peters, H., Theclinical dose response to aspirin, Br. J. Anaesth., 40 (1968) 433–441.

[58] Patel, A., Skelly, A.M., Kohn, H. and Preiskel, H.W., Double-blindplacebo controlled comparison of the analgesic effects of single doses oflornoxicam and aspirin in patients with postoperative dental pain, Br.Dent. J., 170 (1991) 295–299.

[59] Rowe, N.H., Aseltine, L.F. and Turner, J.L., Control of pain withmeclofenamate sodium following removal of an impacted molar, OralSurg. Oral Med. Oral Pathol., 59 (1985) 446–448.

[60] Seymour, R.A., Weldon, M., Kelly, P., Nicholson, E. and Haw-kesford, J.E., An evaluation of buffered aspirin and aspirin tablets in post-operative pain after third molar surgery, Br. J. Clin. Pharmacol., 33 (1992)395–399

[61] Seymour, R.A., Williams, F.M., Luyk, N.M., Boyle, M.A., Whit-field, P.M., Nicholson, E., Booth P. W. and Rawlins, M.D., Comparativeefficacy of soluble aspirin and aspirin tablets in postoperative dental pain,Eur. J. Clin. Pharmacol., 30 (1986) 495–498.

[62] Sunshine, A., Marrero, I., Olson, N.Z., Laska, E.M. and McCor-mick, N., Oral analgesic efficacy of suprofen compared to aspirin, aspirinplus codeine, and placebo in patients with postoperative dental pain, Phar-

296 J.E. Edwards et al. / Pain 81 (1999) 289–297

macology, 27 Supp. (1983) 31–40.

[63] Sunshine, A., Olson, N.Z., Laska, E.M., Zighelboim, I., De Castro,A. and De Sarrazin, C., Analgesic effect of graded doses of flurbiprofen inpost episiotomy pain, Pharmacotherapy, 3 (1983) 177–181.

[64] Sunshine, A., Olson, N.Z., Laska, E.M., Zighelboim, I., DeCastro,A. and DeSarrazin, C., Ibuprofen, zomepirac, aspirin, and placebo in therelief of postepisiotomy pain, Clin. Pharmacol. Ther., 34 (1983) 25–48.

[65] Sunshine, A., Roure, C., Colon, A., Olson, N.Z., Gonzalez, L.,Siegel, C. and Laska, E., Analgesic efficacy of piroxicam in the treatmentof postoperative pain, Am. J. Med., 84 (1988) 16–22.

[66] Wang, R.I. and Waite, E.M., The clinical analgesic efficacy of oralnefopam hydrochloride, J. Clin. Pharmacol., 19 (1979) 395–402.

[67] Wang, R.I.H., Johnson, R.P., Lee, J.C.M. and Waite, E.M., Theoral analgesic efficacy of bicifadine hydrochloride in postoperative pain, J.Clin. Pharmacol., 22 (1982) 160–164.

[68] Winter, L.Jr., Appleby, F., Ciccone, P.E. and Pigeon, J.G., Acomparative study of an acetaminophen analgesic combination and aspirinin the treatment of post-operative oral surgery pain, Curr. Ther. Res. Clin.Exp., 33 (1983) 200–213.

[69] Winter Jr., L. and Post, A., Double blind comparison of single oraldoses of oxaprozin, aspirin, and placebo for relief of post operative oralsurgery pain, T. Int. Med. Res., 11 (1983) 308–314.

References

Brune, K. and Lanz, R., Nonopioid analgesics. In: M. Kuhar, G. Pasternak(Eds.), Analgesics: Neurochemical, Behavioural and Clinical Perspec-tives, Raven Press, New York, 1984, pp. 149–173.

Bruster, S., Jarman, B., Bosanquet, N., Weston, D., Erens, R. andDelbanco, T.L., National survey of hospital patients, Br. Med. J., 309(1994) 1542–1546.

Carstens, M., Ligueros, O., Lobo, R., Pauz, M. and Gallardo, F., Compara-tive double blind study of the analgesic and anti inflammatory activity ofGlucamethacin (Indicin), aspirin and placebo in oral surgery, Odontol.Chil., 35 (1987) 31–39.

Collins, S.L., Moore, A., McQuay H.J. and Wiffen, P.J., Oral ibuprofenand diclofenac in post-operative pain: a quantitative systematic review,Eur. J. Pain, 2 (1998) 285–291.

Collins, S.L., Moore, R.A. and McQuay, H.J., The visual analogue painintensity scale: what is moderate pain in millimetres? Pain, 72 (1997)95–97.

Cook, R.J. and Sackett, D.L., The number needed to treat: a clinicallyuseful measure of treatment effect, Br. Med. J., 310 (1995) 452–454.

Cooper, S.A., Efficacy of zomepirac in oral surgical pain, J. Clin.Pharmacol., 20 (1980) 230–242.

Cooper, S.A., Single-dose analgesic studies: the upside and downside ofassay sensitivity. In: M.B. Max, R.K. Portenoy, E.M. Laska (Eds.), TheDesign of Analgesic Clinical Trials (Advances in Pain Research andTherapy, Vol. 18), Raven Press, New York, 1991, pp. 117–124.

Cooper, S.A., Wagenberg, B., Zissu, J., Kruger, G.O., Reynolds, D.C.,Gallegos, L.T., Friedmann, N. and Darnna, R.P., The analgesic efficacyof suprofen in peridontal and oral surgical pain, Pharmacotherapy, 6(1986) 267–276.

Cordero, R.R., A comparative study of Lagundi and aspirin as analgesicsfor post-extraction pain, J. Phillip. Dent. Assoc., 35 (1985) 15–18.

De Vroey, P., A double blind comparison of diflunisal and aspirin in thetreatment of post operative pain after episiotomy, Curr. Med. Res. Opin.,5 (1978) 544–547.

Edwards, J.E., McQuay, H.J., Moore, R.A. and Collins, S.L., Guidelinesfor reporting adverse effects in clinical trials should be improved –lessons from acute pain, (1999) submitted.

FDA Talk Paper, Wyeth-Ayerst Laboratories announces the Withdrawal ofDuract from the Market, http://www.fda.gov/bbs/topics/ANSWERS/ANS00879.html (22 June 1998).

Gallardo, F., Rossi, E. and Ciscutti, V., Analgesic efficacy of ketoprofenon postoperative pain following periodontal surgery, IRCS Med. Sci., 10(1982) 1036–1037.

Gallardo, R.F., Rossi, V.E. and Lobos, T.M., Indoprofen and aspirin for thecontrol of postoperative pain following periodontal surgery, Acta Odon-tol. Latinoam., 1 (1984) 87–91.

Gardner, M.J. and Altman, D.G., Confidence intervals rather thanPvalues: estimation rather than hypothesis testing, Br. Med. J., 292(1986) 746–750.

Hawkey, C.J., Cullen, D.J., Greenwood, D.C., Wilson, J.V. and Logan,R.F., Prescribing of nonsteroidal anti-inflammatory drugs in generalpractice: determinants and consequences, Aliment. Pharmacol. Ther.,11 (1997) 293–298.

Henry, D., Lim, L.L.Y., Rodriguez, L.A.G., Gutthann, S.P., Carson, J.L.,Griffin, M., Savage, R., Logan, R., Moride, Y., Hawkey, C., Hill, S. andFries, J.T., Variability in risk of gastrointestinal complications withindividual non-steroidal anti-inflammatory drugs: results of a collabora-tive meta-analysis, Br. Med. J., 312 (1996) 1563–1566.

Honig, W.J., The use of diflunisal in post operative pain: a report of doubleblind comparative trials in patients after meniscectomy, Curr. Med. Res.Opin., 5 (1978) 536–543.

Houde, R.W., Wallenstein, S.L. and Beaver, W.T., Clinical measurementof pain. In: G. De Stevens (Ed.), Analgetics, Academic Press, NewYork, 1965, pp. 75–122.

Jadad, A.R., Carroll, D., Moore, A. and McQuay, H., Developing a data-base of published reports of randomized clinical trials in pain research.Pain, 66 (1996a) 239–246.

Jadad, A.R., Moore, R.A., Carroll, D., Jenkinson, C., Reynolds, D.J.M.,Gavaghan, D.J. and McQuay, H.J., Assessing the quality of reports ofrandomized clinical-trials: is blinding necessary? Controlled Clin.Trials, 17 (1996b) 1–12.

Khan, K.S., Daya, S. and Jadad, A.R., The importance of quality of pri-mary studies in producing unbiased systematic reviews, Arch. Intern.Med., 156 (1996) 661–666.

L’Abbe, K.A., Detsky, A.S. and O’ Rourke, K., Meta-analysis in clinicalresearch, Ann. Intern. Med., 107 (1987) 224–233.

Mandujano, V.M., Estudio comparativo del Floctafenina y acido acetisa-licilo en un servico ocorinolaringolgia, Invest. Med. Int., 3(6), 379.

McQuay, H.J., Carroll, D. and Moore, R.A., Injected morphine in post-operative pain; a quantitative systematic review, J. Pain Symptom Man-age., (1999) in press.

McQuay, H.J. and Moore, R.A., An Evidence-Based Resource for PainRelief, Oxford University Press, Oxford, 1998.

Moore, A., McQuay, H. and Gavaghan, D., Deriving dichotomous out-come measures from continuous data in randomized controlled trials ofanalgesics, Pain, 66 (1996) 229–237.

Moore, A., McQuay, H. and Gavaghan, D., Deriving dichotomous out-come measures from continuous data in randomized controlled trials ofanalgesics: verification from independent data, Pain, 69 (1997a) 127–130.

Moore, A., Moore, O., McQuay, H. and Gavaghan, D., Deriving dichot-omous outcome measures from continuous data in randomized con-trolled trials of analgesics: use of pain intensity and visual analoguescales, Pain, 69 (1997b) 311–315.

Moore, A., Collins, S., Carroll, D. and McQuay, H., Paracetamol with andwithout codeine in acute pain: a quantitative systematic review, Pain, 70(1997c) 193–201.

Norholt, S.E., SindetPedersen, S., Larsen, U., Bang, U., Ingerslev, J.,Nielsen, O., Hansen, H.J. and Ersboll, A.K., Pain control after dentalsurgery: a double-blind, randomized trial of lornoxicam versusmorphine, Pain, 67 (1996) 335–343.

Or, S. and Bozkurt, A., Evaluation of the analgesic effects of variousaspirin combinations in oral surgery, Ankara Univ Hekim Fak Derg,12 (1985) 321–332.

Rye, L.A. and Siegel, S.R., A study of factors influencing the qualitativeresponse and/or the time interval of response to mild analgesics forpostoperative pain, Georgetown Dent. J., 62 (1978) 36–41.

297J.E. Edwards et al. / Pain 81 (1999) 289–297