{"paper_id":"52edb40d-c146-421e-aba4-e073b8ee65f6","body_text":"BioMed Central\nPage 1 of 7\n(page number not for citation purposes)\nBMC Women's Health\nOpen AccessResearch article\nRofecoxib for dysmenorrhoea: meta-analysis using individual \npatient data\nJayne E Edwards, R Andrew Moore* and Henry J McQuay\nAddress: Pain Research Unit & Nuffield Department of Anaesthetics University of Oxford The Churchill Headington Oxford OX3 7LJ UK\nEmail: Jayne E Edwards - jayne.edwards@pru.ox.ac.uk; R Andrew Moore* - andrew.moore@pru.ox.ac.uk; \nHenry J McQuay - henry.mcquay@pru.ox.ac.uk\n* Corresponding author    \nAbstract\nBackground: Individual patient meta-ana lysis to determine the anal gesic efficacy and adverse\neffects of single-dose rofecoxib in primary dysmenorrhoea.\nMethods: Individual patient informat ion was available from three randomised, double blind,\nplacebo and active controlled tr ials of rofecoxib. Data were combined through meta-analysis.\nNumber-needed-to-treat (NNT) for at least 50% pa in relief and the proportion of patients who\nhad taken rescue medication over 12 hours were calculated. Information was collected on adverse\neffects.\nResults: For single-dose rofecoxib 50 mg compared with placebo, the NNTs (with 95% CI) for at\nleast 50% pain relief were 3.2 (2.4 to 4.5) at six, 3.1 (2.4 to 9.0) at eight, and 3.7 (2.8 to 5.6) at 12\nhours. For naproxen sodium 550 mg they were 3.1 (2.4 to 4.4) at six, 3.0 (2.3 to 4.2) at eight, and\n3.8 (2.7 to 6.1) at 12 hours. The proportion of patients who needed rescue medication within 12\nhours was 27% with rofecoxib 50 mg, 29% with naproxen sodium 550 mg, and 50% with placebo.\nIn the single-dose trial, the proportion of patients reporting any adverse effect was 8% (4/49) with\nrofecoxib 50 mg, 12% (6/49) with ibuprofen 400 mg, and 6% (3/49) with placebo. In the other two\nmultiple dose trials, the proportion of patients reporting any adverse effect was 23% (42/179) with\nrofecoxib 50 mg, 24% (45/181) with naproxen sodium 550 mg, and 18% (33/178) with placebo.\nConclusions: Single dose rofecoxib 50 mg  provided similar pain relief to naproxen sodium 550\nmg over 12 hours. The duration of analgesia with rofecoxib 50 mg was similar to that of naproxen\nsodium 550 mg. Adverse effects were uncommon suggesting safety in short-term use of rofecoxib\nand naproxen sodium. Future research should in clude restriction on daily life and absence from\nwork or school as outcomes.\nBackground\nDysmenorrhoea is associated with painful cramping of\nthe lower abdominal or back muscles, with or without\nother symptoms such as nausea, vomiting, and diarrhoea.\nOnset of dysmenorrhoea is common during adolescence,\nand up to 50% of women of reproductive age may be\naffected [1], and 10% incapacitated for up to three days\neach menstrual cycle. The pain caused by dysmenorrhoea\ncan be debilitating, resulting in women being unable to\nperform daily activities, and being absent from work or\nPublished: 20 July 2004\nBMC Women's Health 2004, 4:5 doi:10.1186/1472-6874-4-5\nReceived: 23 March 2004\nAccepted: 20 July 2004\nThis article is available from: http://www.biomedcentral.com/1472-6874/4/5\n© 2004 Edwards et al; licensee BioMed Central Ltd. This is an open-access article distributed under the terms of the Creative Commons Attribution \nLicense (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original \nwork is properly cited.\n\nBMC Women's Health 2004, 4:5 http://www.biomedcent ral.com/1472-6874/4/5\nPage 2 of 7\n(page number not for citation purposes)\nschool. In consequence, dysmenorrhoea is associated\nwith emotional, social, and economic burdens. Despite\nthe impact of dysmenorrhoea on daily living, few women\nseek medical advice [2], or know which treatments work\n[3].\nRaised concentrations of uterine prostaglandins are\nthought to cause the pain and cramping associated with\ndysmenorrhoea [4-6]. Nonsteroidal anti-inflammatory\ndrugs (NSAIDs) inhibit prostaglandin synthesis and are\ncommonly used to treat the condition. The newer Cox-2\nselective inhibitors (coxibs) also inhibit prostaglandin\nsynthesis, providing an alternative to conventional\nNSAIDs. Relatively low rates of gastrointestinal adverse\neffects allow the use of higher doses of coxibs in acute\npain and dysmenorrhoea. These high doses may have the\nadditional advantage of longer duration analgesia with\nextended dosing intervals.\nSystematic reviews have shown NSAIDs to be effective in\nthe treatment of primary dysmenorrhoea [7,8] While the\nlatest, Cochrane, review [8] reported on 4,066 women in\ntrials of NSAIDs in dysmenorrhoea, the trials themselves\nwere small, with an average of about 50 women per trial.\nThese 63 randomised double-blind trials investigated 21\ndifferent NSAIDS, at different doses, in studies of varying\ndesign, varying outcomes, and varying duration.\nAt least moderate pain relief over a cycle was reported in\n14 comparisons between NSAID (any NSAID, any dose)\nand placebo in 599 women, with an NNT of 2.1 (1.9 to\n2.5). The most studied NSAID was naproxen, with 287\nwomen in seven trials, with an NNT of 2.5 (2.0 to 3.3) for\nthis outcome. An earlier review [7] included more trials,\nsome not double-blind, but came to substantially the\nsame conclusion about analgesic efficacy in primary dys-\nmenorrhoea. NSAIDs also improved activities of daily liv-\ning, though reported in only 216 women [8]. Adverse\nevent information in these trials was not informative\ngiven the small number of women and trials, and that\nadverse events were rare in younger women taking\nNSAIDs for limited time.\nCox-2 selective inhibitors (coxibs) provide an alternative\nto conventional NSAIDs, with a potential advantage of\nonce daily dosing [9-11]. This individual patient data\nmeta-analysis of rofecoxib in dysmenorrhoea aimed to\ndetermine the efficacy and duration of analgesic activity of\nsingle dose rofecoxib, and to evaluate adverse effects.\nMethods\nQUOROM guidelines [12] for quality of reporting of\nmeta-analyses were followed though no flow chart was\nused since trial data all came from a single source.\nMerck Research Laboratories, Rahway, New Jersey pro-\nvided individual patient data from three Phase III trials of\nrofecoxib in dysmenorrhoea (studies 38, 55 and 56), with\nthe guarantee that all relevant studies completed by July\n2002 had been made available. One of the trials has been\npublished in full [9]. Searching PubMed to January 2004\nidentified one other randomised but open trial of\nrofecoxib [13] not sponsored by Merck. For inclusion, tri-\nals had to be randomised and double blind, compare\nrofecoxib with placebo, and provide single dose efficacy\ninformation. Outcome data were available for pain relief,\npain intensity, time to remedication (use of rescue analge-\nsic), and adverse effects. Each trial report was independ-\nently read and scored for quality using a three item, 0–5\npoint scale [14]. For inclusion a trial had to score a mini-\nmum of two points, one each for randomisation and dou-\nble blinding, out of a maximum of 5. A sixteen-point scale\nwas also used to assess trial validity [15].\nOur intention was to use two pain outcomes (pain relief\nover the first dose, and pain relief over the whole cycle)\nwith the outcome being that closest to at least half pain\nrelief. Over the first dose, this might be a measure of total\npain relief (TOTPAR), while over a whole cycle it may be\na patient global evaluation of good or excellent, rather\nthan mild, fair, no improvement, or worse. Analyses for\ncomparator treatments were based on information availa-\nble only from the trials of rofecoxib in this report. Out-\ncome data were pooled in an intention to treat (number\nof patients randomised) analysis. Neither heterogeneity\ntests nor funnel plots were used [16,17]. Instead clinical\nhomogeneity of trials was examined graphically [18].\nRelative benefit (or risk) was calculated using a fixed\neffects model [19] with no statistically significant differ-\nence between treatments assumed when the 95% confi-\ndence intervals included 1. Number-needed-to-treat (or\nharm) was calculated using the method of Cook and Sack-\nett [20] using pooled observations. NNT is the reciprocal\nof the absolute risk reduction or increase; for instance, if\n75 out of 100 patients benefit with treatment and only 25\nout of 100 benefit with placebo, the absolute risk increase\nis 0.75–0.25 = 0.5, and the NNT is 1/0.5 = 2. The z test\n[21] was used to determine statistical differences between\nNNTs for different doses, treatments or outcomes. Mean\nadverse event rates were calculated, weighting by treat-\nment group size. Use of rescue medication was analysed as\nthe proportion of patients remedicating at different time\npoints within 12 hours.\nResults\nThe mean age of women in the trials was 31 years, and at\nbaseline pain was moderate in 66% of women and severe\nin 34%. All trials were randomised, double blind, and\ncompared oral doses of rofecoxib with an active control\n\nBMC Women's Health 2004, 4:5 http://www.biomedcent ral.com/1472-6874/4/5\nPage 3 of 7\n(page number not for citation purposes)\nand placebo in women with moderate to severe pain due\nto dysmenorrhoea. One trial was a single dose crossover\n(study 38), and two were multiple dose crossovers (stud-\nies 55 and 56), where the crossover was between single\ndoses in different menstrual cycles. Single dose efficacy\ndata were available for the first 12 hours of treatment in\nall trials, but not summary estimates over a cycle. Study\ndesigns and quality and validity scores are shown in Table\n1. All trials scored the maximum five points for quality\nand at least 13/16 points for trial validity; some of the cri-\nteria for validity were not appropriate because of the indi-\nvidual patient presentation of results.\nThe single dose trial (study 38) was conducted over three\ncycles. Each of 49 women received three of four treat-\nments (rofecoxib 25 mg, rofecoxib 50 mg, ibuprofen 400\nmg, or placebo). For the two multiple dose studies, one\n(study 55) compared a single dose of rofecoxib 50 mg fol-\nlowed by 25 mg daily as required, naproxen sodium 550\nmg every 12 hours, or placebo in 60 women over three\nmenstrual cycles. The other (study 56) compared\nrofecoxib 50 mg as required, rofecoxib 50 mg followed by\n25 mg daily as required, naproxen sodium 550 mg every\n12 hours, or placebo in 122 women over four menstrual\ncycles. Both trials reported multiple dose adverse effects.\nIn the multiple dose trials, all women received each treat-\nment regimen for one cycle.\nPain intensity and pain relief were measured using the\nstandard 4-point categorical pain intensity scale (0 none,\n1 mild, 2 moderate, 3 severe) and a 5-point point pain\nrelief scale (0 none, 1 a little, 2 some, 3 a lot, 4 complete).\nPain measurements were collected using patient diaries.\nPatients were assessed at baseline, then at least hourly for\neight hours, and again at 12 hours for single dose efficacy\ndata. The exact time at which a patient requested remedi-\ncation (or rescue analgesic), if required, was recorded.\nAdverse effects were recorded as the number of patients\nwith any adverse effect(s), or particular adverse effects.\nEfficacy\nFull efficacy results over six, eight and 12 hours are shown\nin Table 2. All active treatments were significantly more\neffective than placebo at all time points.\nRofecoxib 25 mg was tested in a single trial, rofecoxib 50\nmg in three, ibuprofen 400 mg in one and naproxen\nsodium 550 mg in two. For all the active analgesics the\nproportion of patients with at least 50% pain relief was\nabout 60% at all time points, and with placebo it was\nabout 30% at all time points (Table 2). Numbers needed\nto treat tended to be much the same for rofecoxib 50 mg,\nibuprofen 400 mg and naproxen sodium 550 mg, though\nsomewhat higher (worse) for rofecoxib 25 mg (Table 2).\nThere was no significant difference between NNTs for sin-\ngle doses of study treatments at six, eight or 12 hours. For\ninstance, no significant difference at the 12 hour compar-\nison was seen between the NNTs of rofecoxib 25 mg and\nrofecoxib 50 mg (z score 1.19, p = 0.23), ibuprofen 400\nmg (z score 0.28, p = 0.78), or naproxen sodium 550 mg(z\nscore 1.09, p = 0.28), or between NNTs of rofecoxib 50 mg\nand ibuprofen 400 mg (z score 0.26, p = 0.79), or\nnaproxen sodium 550 mg (z score 0.096, p = 0.60).\nRemedication\nThe proportion of patients who remedicated at different\ntime points over 12 hours is shown in Figure 1. At 12\nhours remedication occurred with 29% on rofecoxib 25\nmg, 28% on rofecoxib 50 mg, 29% on naproxen sodium\n550 mg, 41% on ibuprofen 400 mg, and 50% with\nplacebo.\nTable 1: Trial details\nTrial ID Study drug and dose, number of \nwomen\nDesign Observations after 8 \nhrs\nQuality \nscore\nValidity \nscore\n38 49 women\nRofecoxib 25 mg\nRofecoxib 50 mg\nIbuprofen 400 mg\nPlacebo\nSingle oral dose, parallel 3 menstrual \ncycles\n12, 24 5/5 ≥13/16\n55 60 women\nRofecoxib 50 mg then 25 mg as required\nNaproxen sodium 550 mg every 12 hrs\nPlacebo\nOral. Multiple dose study with single \ndose efficacy data, multiple dose \nadverse events Cross-over, 1 of 6 \ndrug sequences 3 menstrual cycles\n12 hour obervations after \na single dose in a three-\nday study\n5/5 ≥13/16\n56 122 women\nRofecoxib 50 mg as required\nRofecoxib 50 mg then 25 mg as required\nNaproxen sodium 550 mg every 12 hrs\nPlacebo\nOral. Multiple dose study with single \ndose efficacy data, multiple dose \nadverse events Cross-over, 1 of 4 \ndrug sequences 4 menstrual cycles\n12 hour obervations after \na single dose in a three-\nday study\n5/5 ≥13/16\n\nBMC Women's Health 2004, 4:5 http://www.biomedcent ral.com/1472-6874/4/5\nPage 4 of 7\n(page number not for citation purposes)\nRemedication time for all drugsFigure 1\nRemedication time for all drugs\nTable 2: Number needed to treat for at least 50% pain relief\nImproved with % improved\nNumber of \ntrials\nDrug and \ndose (mg)\nActive Placebo Active Placebo Relative risk \n(95% CI)\nNNT         \n(95% CI)\nSix hour ourcomes\n1 Rofecoxib 25 66/115 45/118 57 38 1.5 (1.1 to 2.0) 5.0 (3.7 to 7.8)\n3 Rofecoxib 50 140/226 70/225 62 31 2.0 (1.6 to 2.5) 3.2 (2.4 to 4.5)\n1 Ibuprofen 400 31/49 10/47 63 21 3.0 (1.7 to 5.4) 2.4 (1.7 to 4.2)\n2 Naproxen \nsodium 550\n120/181 60/178 66 34 2.0 (1.6 to 2.5) 3.1 (2.4 to 4.4)\nEight hour ourcomes\n1 Rofecoxib 25 70/115 44/118 61 37 1.6 (1.2 to 2.2) 4.2 (2.8 to 9.0)\n3 Rofecoxib 50 147/226 73/225 65 32 2.0 (1.6 to 2.5) 3.1 (2.4 to 9.0)\n1 Ibuprofen 400 30/47 11/47 61 21 2.6 (1.5 to 4.6) 2.6 (1.8 to 5.1)\n2 Naproxen \nsodium 550\n121/181 62/178 68 35 2.0 (1.6 to 2.4) 3.0 (2.3 to 4.3)\nTwelve hour ourcomes\n1 Rofecoxib 25 64/115 45/118 56 38 1.5 (1.1 to 1.9) 5.7 (3.3 to 20)\n3 Rofecoxib 50 135/226 74/225 60 33 1.8 (1.5 to 2.3) 3.7 (2.8 to 5.6)\n1 Ibuprofen 400 27/49 12/47 55 26 2.2 (1.3 to 3.7) 3.4 (2.1 to 9.2)\n2 Naproxen \nsodium 550\n111/181 62/178 61 35 1.8 (1.4 to 2.2) 3.8 (2.7 to 6.1)\n68 1 2\n0\n10\n20\n30\n40\n50\nPercent remedicating\nHours after dose\nRofecoxib 25 mg\nRofecoxib 50 mg\nIbuprofen 400 mg\nNaproxen sodium 550 mg\nPlacebo\n\nBMC Women's Health 2004, 4:5 http://www.biomedcent ral.com/1472-6874/4/5\nPage 5 of 7\n(page number not for citation purposes)\nAdverse effects\nFew adverse effects of a particular type were reported, and\nnone were serious in any trial. The most commonly\nreported adverse effects were nausea and somnolence, but\nthese occurred infrequently. A single dose in one trial\n(study 38) gave the proportion of patients reporting any\nadverse effect(s) as 10% (5/49 patients) with rofecoxib 25\nmg, 8% (4/49) with rofecoxib 50 mg, 12% (6/49) with\nibuprofen 400 mg, and 6% (3/49) with placebo. With\nmultiple doses over a cycle, the proportion of patients\nreporting any adverse effect(s) was 23% (42/179 patients)\nwith rofecoxib 50 mg, 24% (45/181) with naproxen\nsodium 550 mg, and 18% (33/178) with placebo.\nDiscussion\nPain with dysmenorrhoea usually lasts for about three\ndays, though with considerable individual variation. Tri-\nals of analgesics can have various forms. The simplest\nmight be to give the same analgesic for the whole of the\npainful cycle, and ask a global question concerning effi-\ncacy at the end. Women might then be crossed over to a\ndifferent treatment at the next cycle. A variation would be\nto use the same basic structure, but make more detailed\nevaluations of pain or pain relief over a limited time dur-\ning the first day, though a global question could always be\nadded. A more complicated design would use a cross-over\nwithin a single cycle.\nThe three trials described here used a cross-over between\ncycles, with detailed pain measurements over 12–24\nhours in the first painful day. Twelve-hour and 24-hour\noutcomes have also been reported in two other recent\nstudies of coxibs in dysmenorrhoea [10,11]. With previ-\nous NSAID studies [8] the outcome most often used in\nplacebo-controlled trials was at least moderate pain relief\nor an equivalent outcome over a whole cycle. A recent\nopen-label study had a crossover design with drugs given\neach successive day [13].\nAll three trials were of the highest reporting quality, and\nhad high validity scores, indicating that known sources of\nbias are unlikely to occur [22,8]. We know that to be sure\nof a result (as an NNT) we need information from about\n400 patients when the NNT is about 2, but much more\nwhen the NNT is higher (worse) [23]. Here we have infor-\nmation from about 450 women with rofecoxib 50 mg and\n360 with naproxen sodium 550 mg, but only about 200\nwomen contributed for rofecoxib 25 mg and fewer than\n100 for ibuprofen 400 mg (Table 2). For rofecoxib 25 mg\nand ibuprofen 400 mg, therefore, uncertainty over the size\nof the effect continues.\nIndividual patient information from three trials of high\nquality showed that for the outcome of at least half pain\nrelief over 12 hours, rofecoxib, naproxen sodium and ibu-\nprofen were similarly effective in the treatment of pain\nassociated with dysmenorrhoea. This confirms what was\nknown from previous meta-analysis [8], in which most\ninformation was for naproxen at various doses with an\nNNT of 2.5 (2.0 to 3.3) for the outcome of at least moder-\nate pain relief over 3–5 days compared with placebo. In\nthis analysis the NNT for a single dose of 550 mg\nnaproxen sodium was between 3.0 and 3.8 over six to 12\nhours. The Cochrane review had information on 287\nwomen in seven placebo-controlled trials, while two of\nthe three trials here had information on 359 women tak-\ning naproxen sodium.\nRofecoxib 50 mg was statistically indistinguishable from\nnaproxen sodium 550 mg (Table 2) at all times, though\nrofecoxib 25 mg tended to have numerically higher\n(worse) NNTs at all times. Remedication over 12 hours\nwas statistically indistinguishable between rofecoxib\ndoses and naproxen. Here the number of women studied\nwas even larger, with 451 women involved in the trials\ncomparing rofecoxib 50 mg and placebo. The difference\nbetween rofecoxib 50 mg and naproxen sodium 550 mg\nwould be in dosing schedules, with once versus twice a\nday dosing.\nPain relief and duration of analgesia are not the only\nissues of importance in dysmenorrhoea. The impact of\ndysmenorrhoea on activities of daily living, disability or\nfunction, and absence from work or school are additional\nfactors to be considered. These outcomes were not\naddressed by the trials for rofecoxib, which were con-\nducted for regulatory purposes. This limits the utility of\nthe information, but trials of other coxibs (also conducted\nfor regulatory purposes) have also concentrated on pain\nrelief and duration of analgesia [10,11].\nFuture trials should examine a range of short-term analge-\nsia and longer outcomes like interference with daily living\nor absence from work or school. The analysis by Zhang\nand colleagues [7] did examine these additional out-\ncomes, and found daily life to be less restricted with\nnaproxen or ibuprofen than with placebo, and fewer\nabsences from work or school to occur with naproxen\nthan with placebo. These outcomes are infrequently\nreported [8], but are likely to be associated with pain, so\ndecreased pain should improve these other outcomes as\nwell. Verification of this assumption with data from high\nquality clinical trials would be welcome, though.\nFuture individual patient analysis of trials in dysmenor-\nrhoea would have the potential to examine issues around\nthe efficacy of analgesics in women with heavy menstrual\nloss, or who use combined oral contraceptive pills. In this\nanalysis information was not available for these analyses,\n\nBMC Women's Health 2004, 4:5 http://www.biomedcent ral.com/1472-6874/4/5\nPage 6 of 7\n(page number not for citation purposes)\nand in any event any sub-groups would probably have\nbeen too small for any definitive answer.\nIn the trials for rofecoxib, information on adverse effects\nwas collected using diaries. Few adverse effects were\nreported to have occurred and none were serious. The\nmost common adverse effects were nausea and somno-\nlence. These and headache have been frequently reported\nwith other coxibs [10,11] and NSAIDs [7,24]. The prob-\nlem when interpreting information on adverse effects,\nthough, is that any symptom can be recorded as an\nadverse event however tenuous its association to the study\ndrug. We cannot be certain whether these symptoms were\ndue to the condition or to the drug.\nConclusions\nBased on information from three trials, a single dose of\nrofecoxib 50 mg is as effective as a single dose of naproxen\nsodium 500 mg in controlling the pain associated with\ndysmenorrhoea, and causes relatively few adverse effects.\nCompeting interests\nRAM has been a consultant for Merck, Sharpe and Dohme\nLtd, UK. RAM, JE and HJM have received lecture fees from\npharmaceutical companies. The authors have received\nresearch support from charities and government sources\nat various times, but no such support was received for this\nwork. Neither author has any direct stock holding in any\npharmaceutical company. The terms of the financial sup-\nport from MSD included freedom for authors to reach\ntheir own conclusions, and an absolute right to publish\nthe results of their research, irrespective of any conclu-\nsions reached. MSD did have the right to view the final\nmanuscript before publication, and did so.\nAuthors' contributions\nJE conducted the analyses, which were checked by RAM.\nAll authors contributed equally to the design, writing and\nreviewing of the paper.\nAcknowledgements\nMerck Research Laboratories, Rahway, New Jersey provided individual \npatient data for use in this review. Financial support was provided by an \nunconditional educational grant from Merck Sharp and Dohme Ltd, UK. \nAdditional support was provided by the Oxford Pain Relief Trust and \nOxford Pain Research funds.\nReferences\n1. Dawood MY: Nonsteroidal anti-inf lammatory drugs and\nchanging attitudes toward dysmenorrhoea. American Journal of\nMedicine 1988, 84:23-29.\n2. Hewison A, van den Akker OB: Dysmenorrhoea, menstrual atti-\ntude and GP consultation. British Journal of Nursing 1996, 5:480-4.\n3. Hillen TI, Grbavac SL, Johnst on PJ, Straton JA, Keogh JM: Primary\ndysmenorrhoea in young Western Australian women: prev-\nalence, impact and knowledge of treatment.  Journal of Adoles-\ncent Health 1999, 25:40-5.\n4. Dawood MY: Nonsteroidal antiinflammatory drugs and repro-\nduction. American Journal of Obstetrics and Gynecology  1993,\n169:1255-65.\n5. Bieglmayer C, Hofer G, Kainz C, Reinthaller A, Kopp B, Janisch H:\nConcentrations of various ar achidonic acid metabolites in\nmenstrual fluid are associated with menstrual pain and are\ninfluenced by hormonal contraceptives.  Gynecological\nEndocrinolology 1995, 9:307-12.\n6. Nigam S, Benedetto C, Zonca M, Leo-Rossberg I, Lubbert H, Ham-\nmerstein J: Increased concentrations of eicosanoids and plate-\nlet-activating factor in me nstrual blood from women with\nprimary dysmenorrhea. Eicosanoids 1991, 4:137-41.\n7. Zhang WY, Li Wan Po A: Efficacy of minor analgesics in primary\ndysmenorrhoea: a systematic review. British Journal of Obstetrics\nand Gynaecology 1998, 105:780-9.\n8. Marjoribanks J, Proctor M, Farquhar C: Nonsteroidal anti-inflam-\nmatory drugs for primary dysmenorrhoea.  Cochrane Database\nof Systematic Reviews 2003, 4:CD001751.\n9. Morrison BW, Daniels SE, Kotey P, Cantu N, Seidenberg B:\nRofecoxib, a specific cyclooxyg enase-2 inhibitor, in primary\ndysmenorrhea: a randomized controlled trial.  Obstetrics &\nGynecology 1999, 94:504-8.\n10. Daniels SE, Talwalker S, Torri S,  Snabes MC, Recker DP, Verburg KM:\nValdecoxib, a cyclooxygenase-2-specific inhibitor, is effective\nin treating primary dysmenorrhea. Obstetrics & Gynecology 2002,\n100:350-8.\n11. Malmstrom K, Kotey P, Cichanow itz N, Daniels S, Desjardins PJ:\nAnalgesic efficacy of etoricox ib in primary dysmenorrhea:\nresults of a randomized, controlled trial.  Gynecology Obstetrics\nInvestigations 2003, 56:65-9.\n12. Moher D, Cook DJ, Eastwood S,  Olkin I, Rennie D, Stroup DF:\nImproving the quality of meta-analyses of randomised con-\ntrolled trials: the QUOROM statement.  Lancet 1999,\n354:1896-1900.\n13. Sahin I, Saracoglu F, Kurban Y, Turkkani B: Dysmenorrhea treat-\nment with a single daily dose of rofecoxib.  International Journal\nof Gynaecology and Obstetrics 2003, 83:285-91.\nTable 3: Proportion of women who used rescue analgesic\nPercent who remedicated by (hrs)\nDrug & dose (mg) 68 1 2\nRofecoxib 25 21 25 28\nRofecoxib 50 22 24 27\nIbuprofen 400 22 35 41\nNaproxen sodium 550 18 24 29\nPlacebo 37 44 50\n\nPublish with BioMed Central   and  every \nscientist can read your work free of charge\n\"BioMed Central will be the most significant development for \ndisseminating the results of biomedical research in our lifetime.\"\nSir Paul Nurse, Cancer Research UK\nYour research papers will be:\navailable free of charge to the entire biomedical community\npeer reviewed and published immediately upon acceptance\ncited in PubMed and archived on PubMed Central \nyours — you keep the copyright\nSubmit your manuscript here:\nhttp://www.biomedcentral.com/info/publishing_adv.asp\nBioMedcentral\nBMC Women's Health 2004, 4:5 http://www.biomedcent ral.com/1472-6874/4/5\nPage 7 of 7\n(page number not for citation purposes)\n14. Jadad AR, More RA, Carroll D, Jenkinson C, Reynolds DJM, Gavaghan\nDJ, McQuay HJ: Assessing the quality of reports of randomised\nclinical trials: is blinding necessary? Controlled Clinical Trials 1996,\n17:1-12.\n15. Smith LA, Oldman AD, McQuay HJ, Moore RA: Teasing apart\nquality and validity in systemati c reviews: an example from\nacupuncture trials in chronic neck and back pain.  Pain 2000,\n86:119-132.\n16. Gavaghan DJ, Moore RA, McQuay HJ: An evaluation of homoge-\nneity tests in meta-analyses in pain using simulations of indi-\nvidual patient data. Pain 2000, 85:415-424.\n17. Tang J-L, Liu JLY: Misleading funnel plot for detection of bias in\nmeta-analysis. Journal of Clinical Epidemiology 2000, 53:477-484.\n18. L'Abbé KA, Detsky AS, O'Rourke K: Meta-analysis in clinical\nresearch. Annals of International Medicine 1987, 107:224-233.\n19. Morris JA, Gardner MJ: Calculating confidence intervals for rel-\native risk, odds ratios and st andardised ratios and rates.  In:\nStatistics with confidence – confiden ce intervals and statistical guidelines\nEdited by: Gardner MJ, Altman DG. London, British Medical Journal\nBooks; 1995:50-63. \n20. Cook RJ, Sackett DL: The number needed to treat: a clinically\nuseful measure of treatment effect. British Medical Journal 1995,\n310:452-454.\n21. Tramèr MR, Reynolds DJ M, Moore RA, McQuay HJ: Impact of cov-\nert duplicate publication on meta-analysis: a case study. Brit-\nish Medical Journal 1997, 315:635-9.\n22. Moore RA, Gavaghan D, Tramer MR, Collins SL, McQuay HJ: Size is\neverything-large amounts of information are needed to\novercome random effects in estimating direction and mag-\nnitude of treatment effects. Pain 1998, 78(3):209-216.\n23. Khan KS, Daya S, Jadad AR: The importance of quality of pri-\nmary studies in producing unbiased systematic reviews. Arch\nIntern Med 1996, 156:661-666.\n24. Milsom I, Minic M, Dawood MY, Akin MD, Spann J, Niland NF, Squire\nRA: Comparison of the efficacy and safety of nonprescription\ndoses of naproxen and naproxen sodium with ibuprofen,\nacetaminophen, and placebo in  the treatment of primary\ndysmenorrhea: a pooled an alysis of five studies.  Clinical\nTherapeutics 2002, 24:1384-400.\nPre-publication history\nThe pre-publication history for this paper can be accessed\nhere:\nhttp://www.biomedcentral.com/1472-6874/4/5/prepub","source_license":"CC0","license_restricted":false}