Non-steroidal anti-inflammatory drugs for heavy bleeding or pain associated with intrauterine-device use

Background Heavy bleeding and pain are the most common reasons why women discontinue intrauterine devices (IUDs). Non‐steroidal anti‐inflammatory drugs, which inhibit prostaglandin synthesis, have been shown to be effective in reducing menstrual bleeding and pain in women without IUDs.

This review summarizes all randomized controlled trials studying use of nonsteroidal anti‐inflammatory drugs for treatment of bleeding or pain associated with IUD use. Trials of prophylactic use of these drugs around the time of IUD insertion were also included. Search methods We searched for relevant trials in PubMed, CENTRAL, POPLINE, and LILACS. We also searched for clinical trials in ClinicalTrials.gov and the International Clinical Trials Registry Platform (ICTRP). We wrote to the authors of trials identified to seek other published or unpublished trials. Selection criteria We included all randomized controlled trials in any language that tested one or more nonsteroidal anti‐inflammatory drugs for treatment or prevention of bleeding or pain associated with IUD insertion or use. Data collection and analysis Two authors independently abstracted data from relevant trials, and we entered data into RevMan for analysis. Main results We found 15 trials from 10 countries; the total number of participants was 2702. Nonsteroidal anti‐inflammatory drugs (naproxen, suprofen, mefenamic acid, ibuprofen, indomethacin, flufenamic acid, alclofenac, and diclofenac) were effective in reducing menstrual blood loss associated with IUD use. This held true for women with and without complaints of heavy bleeding. Similarly, these drugs were effective in reducing pain associated with IUD use. In contrast, prophylactic use of nonsteroidal anti‐inflammatory drugs had mixed results; studies with ibuprofen found no effect on pain after insertion on IUD discontinuation. No important differences emerged in the one trial comparing the effect of different NSAIDs on bleeding. Authors' conclusions Nonsteroidal anti‐inflammatory drugs reduce bleeding and pain associated with IUD use. NSAIDs should be considered first‐line therapy; if NSAIDs are ineffective, tranexamic acid may be considered as second‐line therapy. Prophylactic ibuprofen administration with the first six menses after insertion appears unwarranted. Plain language summary Pain relievers for bleeding and pain related to intrauterine devices used for birth control Heavy menstrual bleeding and cramping are the most common reasons why women stop using an intrauterine device (IUD) for birth control. A class of drugs (nonsteroidal anti‐inflammatory drugs, or NSAIDs) reduces menstrual bleeding and cramping in women who are not using an IUD. These drugs, such as naproxen and ibuprofen, are sold over‐the‐counter as pain relievers in many countries. Hence, researchers have studied whether these same drugs might reduce bleeding and pain associated with use of intrauterine devices. This might lead to more comfortable use of IUDs for longer periods of time. We searched for and summarized all the randomized controlled trials that looked at using these drugs to treat bleeding or pain related to an IUD. We also included trials that studied the use of these drugs to prevent these problems. We found 15 trials from 10 countries, with more than 2700 women studied. These drugs reduced both bleeding and pain with intrauterine device use. Whether one drug is better than another was not clear. Similarly, the best dosing was not clear. Preventive treatment with these drugs around the time of IUD insertion had mixed results. No serious problems were reported, but stomach upset and sleepiness can occur with this class of drugs. Because of their safety, low cost, and wide availability, these drugs are appropriate treatment for women who have troublesome bleeding or pain with IUD use.

Abnormal uterine bleeding and pain are the most common medical reasons for premature discontinuation of the intrauterine device (IUD). The IUD is the most common method of reversible contraception worldwide (147 million current users) (United Nations 2003), so premature discontinuation affects large numbers of women. Each year, an estimated 40 million women have an IUD inserted. From 5% to 15% of women discontinue IUD use within one year because of bleeding or pain (Speroff 1996). In a large trial sponsored by the World Health Organization, the cumulative net probability of removal of a Copper T380A IUD for pain or bleeding over 12 years of use was 36 per 100 women (WHO 1997). The corresponding figure with the Copper TCu220C was 34 per 100 women. Both non‐medicated and copper‐bearing devices increase menstrual blood loss. Early studies found that non‐medicated devices, such as the Lippes Loop, caused more bleeding than did copper devices, such as the Copper 7 and Copper T (Gallegost 1978; Guillebaud 1976; Israel 1974). More recent studies with the Copper T380A indicated a persistent increase in menstrual blood loss of about 55%. However, in Swedish women this did not lead to iron deficiency anemia (Milsom 1995). Some women view the increased bleeding as an annoyance. For others, increased bleeding may be unacceptable (Connell 2002). Although cramping is common in the early days after IUD insertion, this commonly resolves (Connell 2002; Speroff 1996). However, some women have an increase in dysmenorrhea (painful, difficult menses) with non‐medicated and copper IUDs. In contrast, progestin‐releasing devices alleviate dysmenorrhea for most women (Pakarinen 2001). Excessive prostaglandin release in the endometrial cavity appears to play a role in both bleeding and pain related to IUDs (Ylikorkala 1994). For example, endometrial concentrations of some prostaglandin metabolites correlate with measured blood loss in IUD users. Prostacyclin (PGI2) causes vasodilation and inhibits platelet aggregation. Thromboxane exists in two forms: A2 is active but unstable and is quickly converted into B2, which is inactive. Thromboxane causes vasoconstriction and blood clotting. Among Chinese women using a copper T IUD and having bleeding greater than 80 mL per cycle, endometrial levels of 6‐keto‐PGF1α (a metabolite of prostacyclin) were significantly higher than those of women with either a levonorgestrel IUD or no IUD (Zhang 1992). In contrast, the endometrial thromboxane B2 concentrations were significantly higher in women using the levonorgestrel IUD. The authors speculated that the imbalance between the prostacyclin and thromboxane A2 may play a role in heavy bleeding with IUDs. Another Chinese study correlated increases in 6‐keto‐PGF1α with menstrual blood loss in IUD users (Zheng 1988). Non‐steroidal anti‐inflammatory drugs (NSAIDs) are prostaglandin synthetase inhibitors. By decreasing production of endometrial prostaglandins, they can alleviate both heavy uterine bleeding (Lethaby 2007) and dysmenorrhea (Marjoribanks 2003). Since the discovery of this effect of NSAIDS in the mid‐1970s, several different drugs in this class have been used to treat or prevent heavy uterine bleeding and pain associated with IUD use as well. This review summarizes the randomized controlled trials evaluating an NSAID for these purposes.

To review all randomized controlled trials that have evaluated an NSAID for treatment or prevention of heavy uterine bleeding or pain associated with IUD use.

Criteria for considering studies for this review Types of studies We included all randomized controlled trials in any language that compared any NSAID for treating heavy uterine bleeding or pain associated with any IUD use. We excluded cohort studies and before‐after studies. Interventions for pain during IUD insertion have been examined in another review (Allen 2009). Types of participants We included women of reproductive age of any gravidity and parity who were using an IUD. Types of interventions Any NSAID used for the treatment or prevention of heavy uterine bleeding or pain associated with IUD was included. Comparisons included a placebo, another NSAID, or other treatment, such as tranexamic acid or desmopressin. Types of outcome measures Outcome measures for bleeding included quantitative measurements of blood loss (Hallberg 1964), semi‐quantitative assessments of blood loss by pictorial blood assessment scores (Higham 1990; Reid 2000), or IUD discontinuation (Hubacher 2006). Participant perceptions of bleeding using categorical scales or menstrual diary cards could also be included. Level of pain was reported on several different Likert scales, ranging from none to extreme pain. We also examined harms of therapy when provided in trial reports. Search methods for identification of studies Electronic searches We searched the computerized databases of MEDLINE via PubMed, POPLINE, Cochrane Central Register of Controlled Trials (CENTRAL), and LILACS. In addition, we searched for recent clinical trials through ClinicalTrials.gov and the International Clinical Trials Registry Platform (ICTRP). The strategies are given in Appendix 1. Strategies for earlier versions of this review can be found in Appendix 2. Searching other resources We wrote to authors of published trial reports to solicit other published or unpublished trials that we may have missed. We also contacted trial report authors as needed to supplement published information. Data collection and analysis We assessed all titles and abstracts found for inclusion. Data extraction and management Two authors (Grimes and Lopez) independently abstracted data from the studies identified to improve accuracy. One author entered data into RevMan 4.2, and a second confirmed correct data entry. Assessment of risk of bias in included studies We evaluated the methodological quality of the trials for potential biases by qualitatively assessing the study design, randomization method, allocation concealment, blinding, premature discontinuation rates, and loss to follow‐up rates. Data synthesis Peto odds ratios with 95% confidence intervals were used for dichotomous outcomes, such as pregnancy. We reported weighted mean differences for measured blood loss. We did not perform subgroup analyses. Since no important 'carryover' effect (Elbourne 2002; Higgins 2005) of NSAIDs has been documented from one menstrual cycle to another (Lethaby 2007; Roy 1981), we combined all measurements from identical treatments in crossover trials. None of these trials was amenable to meta‐analysis (Higgins 2005). Sensitivity analysis Sensitivity analyses were not performed, because we were unable to aggregate these trials.

Description of studies We identified 15 randomized controlled trials that met our eligibility criteria. The total number of participants was 2702, although the number ranged widely in the individual trials, from 10 (Hahn 1979) to 2019 participants (Hubacher 2006). Most trials were small: 12 had fewer than 100, and two had between 100 and 200 participants. The trials came from 10 countries in Europe, the Middle East, North and South America, and Asia. Seven trials focused on prevention, while eight focused on treatment. The prevention trials gave NSAIDs to asymptomatic women as prophylaxis. The treatment trials examined use of NSAIDs as therapy for heavy bleeding, pain, or both. Among the trials studying bleeding, some used quantitative (Hallberg 1964) and others qualitative assessments of blood loss, such as pictorial scores (Higham 1990; Reid 2000) or Likert scales. Prevention trials Seven trials examined prophylactic use of NSAIDs (Hahn 1979; Hubacher 2006; Jensen 1998; Massey 1974; Roy 1981; Toppozada 1982; Ylikorkala 1978). Two of these studied bleeding (Hahn 1979; Roy 1981), two focused on pain (Jensen 1998; Massey 1974), and three studied both of these outcomes (Hubacher 2006; Toppozada 1982; Ylikorkala 1978) Bleeding Hahn 1979 studied the effects of aspirin and paracetamol on menstrual bleeding in a group of 37 Swedish women, 10 of whom were using copper IUDs. Women did not have to have heavy bleeding for inclusion. Each woman received aspirin, paracetamol, or placebo for one month; women then crossed over to alternative treatments, so that each women received each regimen for one month. The outcome measure was menstrual blood loss as determined by the alkaline hematin method (Hallberg 1964). Paracetamol, also known as acetaminophen, has no anti‐inflammatory properties and, unlike aspirin, is not an NSAID (Mayo Clinic 2005). Paracetamol will not be considered further. A trial from Los Angeles examined the impact of ibuprofen on IUD users without heavy bleeding or other complaints (Roy 1981). The researchers recruited 20 women, 10 using a copper IUD and 10 using a Lippes Loop. Participants kept menstrual diaries and collected sanitary products for blood‐loss measurement by the alkaline hematin method. Each woman provided three months of observation: one baseline month, one month using either ibuprofen 1600 mg per day or placebo, and one month crossed over to the alternative treatment. The ibuprofen or placebo was begun at the onset of bleeding and continued until the end of bleeding, or a maximum of seven days. Pain A Danish trial evaluated ibuprofen 600 mg given before IUD insertion, four to six hours after insertion, then again within the first three days (Jensen 1998). Fifty‐five women having either a Nova T or a Gyne T 380 Slimline were randomized to receive either the ibuprofen or a placebo. A 10‐point Likert scale for rating pain was used; one was the least pain and ten was the most intense. An early U.S. trial compared prophylactic administration of naproxen with placebo for Dalkon Shield insertion (Massey 1974). Fifty women, mainly nulliparous (48), were enrolled and allocated to either naproxen or placebo. Treatment began at 2200 hours the evening before the insertion and continued with repeat doses of 300 mg 1.5 hours before insertion, two hours after insertion, and six hours after insertion. Paracervical block with lidocaine was routinely used, given the wide diameter of this device (24 mm) and the nulliparous status of most participants. Outcomes included the need for additional analgesia as well as a complex pain score (including a five‐point Likert scale) developed for this trial. Both bleeding and pain A recent trial from Chile examined the use of prophylactic ibuprofen after insertion of a copper T380A IUD (Hubacher 2006). This large trial enrolled 2019 women in Santiago, Chile, who had never had an IUD before. They were randomized to receive either ibuprofen 1200 mg per day in divided doses or placebo for a maximum of five days with each menses. This regimen was to continue for six menstrual cycles after IUD insertion. The primary outcome measure was removal of the IUD by the 52‐week follow up for any reason, although information was collected to document reasons due to both bleeding and pain. A trial from Egypt compared three different NSAIDs in copper IUD users without a complaint of heavy bleeding (Toppozada 1982). Eighteen parous IUD users were recruited for the four‐month study. Six women were randomized to three different treatment arms: indomethacin, flufenamic acid, or alclofenac. Half of the women in each treatment arm began with the active drug, and the other half started on a placebo. After two months, each woman was crossed over to the alternative placebo or assigned NSAID. Thus, each of the 18 participants used an NSAID for two months and a placebo for two months. Participants maintained diary cards to record bleeding and pain. An early prevention trial was conducted in Finland (Ylikorkala 1978). In this trial, 160 women wanting to have a copper T200 inserted were enrolled in a trial comparing an NSAID (tolfenamic acid) with placebo. At the time of IUD insertion, participants were requested to begin taking either tolfenamic acid 600 mg per day in divided doses or a placebo in the same regimen. Treatment was to continue for seven days, and participants were requested to repeat this regimen starting with menses for the next three cycles. The outcomes of interest included both pain and bleeding, although blood loss was not quantified. Participants and investigators were kept unaware of treatment assignment by use of a placebo, with distribution of drug and placebo in coded vials. Treatment trials Eight trials evaluated NSAIDs as treatment of bleeding, pain, or both (Buttram 1979; Davies 1981; Di Lieto 1987; Lalos 1983; Mercorio 2003; Wu 2000; Yarkoni 1984; Ylikorkala 1983). Four examined bleeding (Davies 1981; Mercorio 2003; Wu 2000; Yarkoni 1984), two studied pain (Buttram 1979; Lalos 1983), and two reported on both bleeding and pain (Di Lieto 1987; Ylikorkala 1983). Bleeding Davies 1981 performed a randomized controlled trial with 48 participants in the U.K. Women had to have documented heavy bleeding (an average of at least 80 mL per cycle on two pre‐treatment cycles) and a normal pelvic examination. The trial had three treatment arms: two different regimens of naproxen and a placebo. The higher naproxen regimen was 1250 mg daily in divided doses; the lower regimen was 1000 mg on the first day and 750 mg daily thereafter. Each woman received one of the treatments above for two menstrual cycles then switched to another regimen for two cycles. Hence, each participant used two of the three regimens over a total of four treatment cycles. The IUDs being used by the 34 participants analyzed included copper devices (12) and non‐medicated devices (22); the latter included Saf‐T‐Coil (11), Lippes Loop (six), and Dalkon Shield (five) IUDs. A recent Italian trial (Mercorio 2003) compared a standard NSAID (mefenamic acid) to a newer treatment: desmopressin acetate. Desmopressin is an analog of vasopressin with longer activity. This drug enhances blood coagulation by mediating the release of von Willibrand factor from endogenous storage sites; this increases circulating Factor VIII and platelet adhesiveness. It also induces uterine contractions because of its similarity to oxytocin. The trial (Mercorio 2003) randomized 24 women judged by pictorial blood assessment scores (Higham 1990) to have bleeding more than 80 mL per cycle. All were wearing a copper IUD. Half the participants received desmopressin (300 mcg) as a nasal spray each morning for five days, starting on the first day of bleeding. Those assigned to mefenamic acid took 1500 mg per day in divided doses for five days, starting on the first day of bleeding. Participants were then followed for three months, with pictorial blood assessment scores as the primary outcome. A Chinese trial compared indomethacin with a traditional Chinese medicine (Wu 2000). Women using copper IUDs and with complaints of heavy or prolonged bleeding were assigned to receive either indomethacin 50 mg per day in divided doses or Baofuxin, a granular preparation of plant products, 16 g per day in divided doses (dissolved in boiling water). The treatments started on the first day of bleeding and continued for seven and 10 days, respectively. Participants kept menstrual diaries for 90 days before the three‐cycle treatment phase and for another 90 days after treatment ended. The outcome was menstrual bleeding patterns as recorded on menstrual diaries; neither bleeding nor spotting was defined. An early trial from Israel (Yarkoni 1984) compared indomethacin 100 mg per day in divided doses versus placebo. Women with perceived heavy bleeding were invited to take part in baseline observations for a two‐month period, during which menstrual blood loss was determined by the alkaline hematin method. Fifteen women had mean blood loss 100 mL or greater, and these were enrolled in the treatment phase. Participants were randomized to receive either indomethacin or placebo; a crossover to the other treatment occurred after two months. Over the course of the four‐month trial, each participant received two months of each treatment. Pain An early U.S. trial (Buttram 1979) randomized 35 women with new or worsened uterine pain associated with IUD insertion. The type of IUDs was not specified. Participants were allocated to receive either naproxen sodium (550 mg initially, then 275 mg every six hours as needed) versus a placebo (lactose) for three consecutive menstrual cycles. The primary outcome of interest was pain, assessed by a self‐administered six‐point Likert scale, with six being excellent relief and one being no relief. A Swedish trial compared naproxen with placebo in a crossover trial (Lalos 1983). Twenty‐one women with primary dysmenorrhea and who noted increased pain after insertion of a copper IUD (either copper T or Cu‐7) were enrolled. They were randomized to receive either naproxen or placebo for two menstrual cycles, after which they crossed over to receive the alternative treatment. The naproxen regimen was 500 mg initially and then 250 mg two to four times a day as needed. Follow up was sought after the second and fourth menstrual periods. Pain was assessed with a five‐point Likert scale, and participants reported subjectively on their bleeding patterns. Both bleeding and pain Di Lieto and colleagues in Italy reported a "double blind crossover study" among 28 women with reported "increased menstrual blood loss and pelvic pain" associated with IUD use (Di Lieto 1987). The trial compared suprofen and placebo. Each woman had one pre‐treatment month of observation, one month on the active drug or placebo, then the final month on the alternative treatment. In this crossover design, each participant took the active drug for a month and the placebo for a month. The IUDs used by participants included the Gravigard (Copper 7) in 18 and the copper T in 10. No quantitative or semi‐quantitative assessments of bleeding were used. Another trial from Finland (Ylikorkala 1983) compared an antifibrinolytic drug (tranexamic acid) versus an NSAID (diclofenac sodium) versus a placebo. Thirty‐one women complaining of heavy bleeding related to IUD use were screened by quantitative measurement of menstrual blood loss by the alkaline hematin method. Twenty‐one were found to have loss greater than 70 mL, and 17 of these qualified as having menorrhagia (greater than 80 mL). Two participants were excluded from analysis (one had leiomyomas, and the other had her IUD removed during the trial). Participants were studied over five cycles: two months on each active treatment and one month on the placebo. The dose of tranexamic acid was 4.5 g per day in divided doses, starting on the first day of bleeding and continuing for five days. The dose of diclofenac sodium started with 150 mg on day one of bleeding and 75 mg per day on days two through five. Placebo was also given for the first five days of bleeding. Risk of bias in included studies Prevention trials Bleeding Hahn 1979 did not explicitly state that participants were allocated by a random method. We infer this from the description of "double‐blind crossover technique." Correspondence with the author indicated that allocation concealment was ensured by using sealed envelopes. The adequacy of treatment blinding with the placebo is unclear. Four of 37 (11%) of women enrolled in the study dropped out: two because of pregnancy and two for unspecified reasons. The sample size was one of convenience. Correspondence with the author indicated that the Los Angeles trial (Roy 1981) used a computer‐generated randomization scheme. Allocation concealment was provided by pharmacy distribution of the drug and placebo. Sample size was one of convenience. A placebo was used to keep participants and investigators unaware of treatment. No losses to follow up were reported. Pain The Danish trial (Jensen 1998) of prophylactic ibuprofen around the time of insertion featured a computer‐generated randomization scheme. Correspondence with the author revealed that allocation concealment was assured by having pharmacy‐generated envelopes containing pills. A formal sample size calculation was done in advance. Participants had a stratified randomization by type of IUD inserted. Blinding was provided by use of a placebo. No participants were lost to follow up. The U.S. trial of naproxen (Massey 1974) featured permuted block randomization with a block size of 10. Allocation concealment was not described. The allocation to method apparently occurred some time before the IUD insertion, since two women assigned to naproxen subsequently changed their minds and did not have the insertion done. Blinding of treatment appears to have been adequate, with identical‐appearing capsules provided in sealed strips. The sample‐size calculation was not provided. All participants who requested additional analgesia were dropped from the trial at that point, leading to considerable missing data. Both bleeding and pain The trial from Chile (Hubacher 2006) was both the largest and most rigorous to date. Randomization was by random permuted blocks, and central pharmacy distribution ensured allocation concealment. Blinding of participants and investigators was provided by using an identical‐appearing placebo. The sample size was determined in advance according to the anticipated reduction in IUD discontinuations by 52 weeks. Losses after randomization were low: 95% of women provided follow‐up information at 52 weeks. Protocol violations were analyzed according to intention‐to‐treat analysis. More detailed analyses of bleeding and pain are in progress; only discontinuation is reported here as an outcome. The Egyptian trial (Toppozada 1982) described randomization by computer. Communication with the author indicated that the sequence was generated by the World Health Organization. WHO trials of that era generally provided allocation concealment by using envelopes that were opaque, sealed, and sequentially numbered. The sample size selection was not described. No losses to follow up were reported. A placebo was used to keep participants and investigators unaware of the treatment assignments. The Finnish prophylactic NSAID trial (Ylikorkala 1978) did not describe the randomization method, allocation concealment, or sample size calculation. Blinding of participants and investigators as to treatment regimen was done by having the drug and placebo distributed in coded vials; whether the blinding was successful is unclear. Due to noncompliance, many participants were excluded from analysis after randomization. For example, 160 women were enrolled and randomized. However, only 122 took the drugs at insertion and with the next menses as requested. By the second menses, only 103 complied with the regimen. Only 85 took the assigned drugs at the third menses. Because of this attrition, determining the number of women portrayed in the results is difficult. The analysis used cycles, rather than participants, as the unit of measurement for some statistical tests, which was improper. Treatment trials Bleeding The U.K. trial (Davies 1981) had a small sample size of 48 and a complex design, with three treatment arms and a crossover after two cycles. This led to data so sparse for some comparisons some planned analyses were abandoned. Fourteen of 48 (29%) of participants were lost or excluded after randomization. The method of randomization was not mentioned, and allocation concealment was not described. Although placebos were used to enable blinding of participants and investigators, whether this worked is unclear from the report, due to difference in numbers of pills taken. The trial in Italy (Mercorio 2003) reported randomization by "computer analysis" and allocation concealment by "envelopes." No sample size calculation was provided. No losses after randomization were reported. Attempts to contact the lead author were unsuccessful. The Chinese trial (Wu 2000) reported that it was "double blind," yet no mention was made of a double dummy or other way to conceal the disparate treatments (one a bag of herbs, the other a capsule). The method of randomization was not specified, and no mention was made of allocation concealment. The sample size calculation was not provided. The lack of blinding likely biased the acceptability assessments of the treatments. Important losses to follow up after randomization appear likely; 90 participants were recruited to each treatment, but only 73 women in the Baofuxin group and 71 in the indomethacin group provided diary information. However, the report indicates that only one participant assigned to Baofuxin was lost to follow up. The Israeli trial of indomethacin (Yarkoni 1984) had no mention of randomization method, description of allocation concealment, or justification for sample size. Moreover, a large proportion of subjects discontinued the trial prematurely: of 15 enrolled, only 9 completed the planned observations. The report did not provide standard deviations for the mean blood losses observed, so these data were not entered into RevMan. Pain The U.S. trial of naproxen sodium (Buttram 1979) for pain did not describe the method of randomization or allocation concealment. The sample size of 35 was not explained. Of 35 women initially enrolled, two dropped out for personal reasons. Six others did not use the assigned treatment for each of three menstrual cycles: three were unsatisfied with the pain relief provided, two had insufficient pain to warrant treatment, and one had her IUD removed. Overall, 23% of participants did not complete the trial as planned. Pain at baseline differed for the two treatment groups; those assigned to the placebo had less pain at baseline that did those assigned to the NSAID (P = 0.02), raising questions about selection bias. Blinding of the treatments was accomplished by use of a placebo; whether the placebo tablets were identical in appearance and taste is unclear. The Swedish study of naproxen featured a placebo treatment and crossover design (Lalos 1983). Correspondence with the author indicated that randomization was done with a table of random numbers. Allocation concealment was not described. Blinding was achieved by distributing the drug and identical‐appearing placebo in amber bottles. A sample size calculation was not provided. Although the trial had approval of an institutional review board, informed consent was verbal rather than written. One participant was inappropriately dropped from analysis because of an intervening operation. Primary presentation of data was in a figure, and raw data were no longer available from the author. Both bleeding and pain Di Lieto 1987 did not explicitly state that they used random allocation to assign treatments, although we inferred from other descriptions that this was a randomized controlled trial. No method of allocation concealment was described. No sample size calculation was provided, and no objective measurement of bleeding was done. Of note, no losses of participants occurred. Attempts to contact the author were unsuccessful. The Finnish trial (Ylikorkala 1983) comparing tranexamic acid and diclofenac versus placebo did not describe the method of randomization or when it was done in the enrollment sequence. No mention was made of allocation concealment or sample size determination. Two participants were apparently dropped after randomization. Otherwise, all 19 participants had complete data. One woman who discontinued diclofenac sodium because of edema was apparently kept in the group to which she was assigned. A double‐dummy approach was used to maintain blinding as to treatment. Effects of interventions Prevention trials Bleeding Aspirin at a daily dose of 1500 mg had no notable effect on menstrual blood loss in asymptomatic copper‐IUD users (Hahn 1979). In this small trial, mean menstrual blood loss for those receiving aspirin was 49 mL and for those receiving placebo 51 mL. In contrast, use of ibuprofen in asymptomatic IUD users reduced menstrual bleeding (Roy 1981). The effect was stronger with the non‐medicated Lippes Loop: mean menstrual blood loss was 30 mL less during ibuprofen use than during placebo use. For women using a copper device, the difference in bleeding was only 10 mL. No serious complications were reported, but one participant using ibuprofen developed edema around her mouth and eye, which resolved after drug discontinuation. Two women reported stomach cramps: one with ibuprofen and one with placebo. Another reported headache, dizziness, nausea, and hot flashes associated with placebo. Pain In the Danish trial, administration of ibuprofen around the time of IUD insertion had no effect on pain (Jensen 1998). Participants having three doses of ibuprofen 600 mg had median pain scores that were not significantly lower than those of women who received placebo; this held true at each of the three assessments. The median pain scores for the ibuprofen and placebo groups were 3.3 and 2.5 at insertion, 1.7 and 1.8 during the first four to six hours, and 1.1 and 1.1 on the third day after insertion, respectively. No important differences were seen in pain by type of IUD. In an early U.S. trial, naproxen given around the time of Dalkon Shield insertion was more effective than placebo in managing pain (Massey 1974). Only seven of 24 women who received naproxen needed additional medication, in contrast to 17 of 26 assigned to placebo. Other analyses were not possible due to missing data. No "untoward effects" or laboratory abnormalities were related to administration of naproxen. Both bleeding and pain Prophylactic administration of ibuprofen 1200 mg per day with the first six menses after IUD insertion had no impact on discontinuation rates (Hubacher 2006). For all reasons combined, 257 of 985 women assigned to ibuprofen discontinued IUD use with 52 weeks, compared with 246 of 977 who took placebo. These figures include spontaneous expulsion (113 and 125, respectively) and removals because of any pain or bleeding (112 and 91, respectively). At 26 weeks, removals for dysmenorrhea or increased menstrual blood loss were 59 of 985 in the ibuprofen group and 61 of 977 women in the placebo group. These removals were related to problems during menses, which is when the women were taking the medication or placebo. The two groups were similar. In asymptomatic Egyptian IUD users, each NSAID tested reduced measured blood loss as compared with a placebo (Toppozada 1982). The mean total blood loss over two months of placebo use and two months of indomethacin was 102.8 and 66.3 mL, respectively. The mean total blood loss over two months of placebo and two months of flufenamic acid was 101.1 and 59.6 mL, respectively. With alclofenac, the corresponding figures were 110.0 and 68.5 mL, respectively. Standard deviations for these means were not provided in the report. However, the authors aggregated all three groups versus placebo and provided standard deviations for these means, allowing for data entry in RevMan. All three NSAIDs decreased total bleeding as compared with placebo; differences between the NSAIDs were minor. Treatment did not shorten the duration of bleeding. Side effects were generally mild; four women assigned to indomethacin had complaints (including headache, nausea, and epigastric pain), one assigned to flufenamic acid had a headache, and two assigned to alclofenac had complaints (intestinal colic and dizziness). Pain was much lower during treatment cycles. Prophylactic administration of tolfenamic acid (Ylikorkala 1978) offered subjective benefit over a placebo. Bleeding immediately after IUD insertion was shorter by about a day, and fewer sanitary pads were required for protection. With subsequent menses, about 15% fewer cycles were deemed heavier than normal with the active drug, and about one pad fewer on average was required. Clots and pain were both less frequent with the NSAID than with placebo. Side effects were noted in both treatment groups. Dyspepsia and diarrhea were reported by four women taking tolfenamic acid, and five taking placebo noted headache. Three participants taking placebo complained of depression and fatigue. Discontinuations because of side effects occurred in three participants in the tolfenamic acid group and two in the placebo group. Treatment trials Bleeding Among women with documented heavy bleeding, naproxen in either of two regimens tested reduced measured blood loss as compared with the placebo (Davies 1981). In addition, the pooled results indicated a decline in blood loss from pre‐treatment cycles with both doses of naproxen. The pooled mean pre‐treatment blood loss was 127 mL; with the higher dose of naproxen, 93 mL; with the lower dose of naproxen, 89 mL; and with placebo, 129 mL. The one patient who dropped out of the trial because of nausea and vomiting of six day's duration had been taking the placebo. No other side effects were reported. Individual responses to naproxen varied widely; some women experienced a 50% reduction in bleeding, while others had no benefit. Mefenamic acid was comparable to desmopressin in reducing heaving bleeding associated with copper‐IUD use (Mercorio 2003). Both treatments led to similar reductions (46% and 41%, respectively) in bleeding scores, compared with baseline. Treatment responses were variable, however. All participants given desmopressin had reduced bleeding; two women assigned to mefenamic acid had a negligible change in bleeding. No serious complications occurred, but three of 12 women assigned to desmopressin reported headache and insomnia. The Chinese trial found indomethacin to be comparable to Baofuxin in reducing number of days of bleeding (Wu 2000). During the three‐cycle treatment phase, the numbers of bleeding/spotting days were similar (mean 28.8 and 28.3, respectively). Likewise, the numbers of bleeding/spotting episodes were comparable (mean 3.4 and 3.3, respectively). The report noted more bleeding/spotting days after treatment with indomethacin than after Baofuxin (mean 30.2 versus 24.7, respectively). A total of 13 women assigned to indomethacin and 12 assigned to Baofuxin had their IUDs removed during the trial because of bleeding, and two others in the indomethacin group had their devices removed for other reasons, such as cramping or pregnancy. The trial from Israel (Yarkoni 1984) found that indomethacin reduced menstrual bleeding by more than 50%. In the baseline cycles, the mean measured blood loss was 174.7 mL. In the placebo cycles, the mean was 159.9. In contrast, during treatment with indomethacin, the mean fell to 75.1 mL, a 57% reduction from baseline, which was statistically significant. No side effects were attributed to indomethacin or placebo. Pain Naproxen sodium improved uterine pain associated with IUD use better than did placebo (Buttram 1979). Of 17 women who took naproxen sodium in one or more cycles, 14 reported "good" or "excellent" pain relief in at least one cycle. In contrast, of 16 participants who took placebo in one or more cycles, only 4 noted "good" or "excellent" relief in one or more cycles. Daily pain relief scores for the 27 women who completed all three cycles confirmed better pain relief with naproxen sodium. The highest relief score possible was 18 (Likert scale of six, added for each of three observation cycles). With naproxen, the mean relief score was 13.9 compared with 10.8 with placebo (P = 0.03). Side effects or harms were not mentioned. The Swedish trial of naproxen (Lalos 1983) found it superior to placebo for pain relief. Of the 20 women with pain assessments, 17 reported better pain relief with the NSAID than with the placebo; only three reported superior relief with the placebo. Calculation of mean pain scores for the two treatments was not possible from published data. Subjective reports of bleeding patterns revealed little impact of naproxen on either amount or duration of flow. Side effects were similar for the two treatments: seven women reported side effects with naproxen, and five did with placebo. None was serious, and none led to trial discontinuation. Gastrointestinal side effects were noted in two and one women, respectively. Both bleeding and pain Suprofen had powerful benefit on both bleeding and pain (Di Lieto 1987). Twenty‐two of 28 women given this NSAID had moderate to "intense" decreases in bleeding, in contrast to none given the placebo. Similarly, with pain, 23 of 28 had moderate to "intense" relief of pain with suprofen, in contrast to only one of 28 with placebo. Three participants reported side effects (stomach cramps, one; nausea and headaches, two). The latter two participants were taking suprofen, but the regimen is not clear for the participant with stomach cramps. The Finnish trial (Ylikorkala 1983) found the antifibrinolytic drug more effective than either the NSAID or placebo in reducing bleeding. The reduction in blood loss was more than 50% with tranexamic acid, compared with only about 20% with diclofenac sodium. Side effects varied by treatment. The most common complaints with tranexamic acid were gastrointestinal disturbances, headache, and sweating. Complaints with diclofenac sodium included sweating, fatigue, dry mouth, headache, itching, and edema; one participant discontinued treatment on the fourth day due to leg edema. Neither drug influenced the duration of bleeding or pelvic pain associated with menses.

Prevention trials NSAIDS reduced bleeding in asymptomatic IUD users. Alclofenac, indomethacin, flufenamic acid, and tolfenamic acid all reduced bleeding when compared with a placebo (Toppozada 1982; Ylikorkala 1978). In one trial, the reduction in bleeding was greater with non‐medicated than with copper devices (Roy 1981). In contrast, aspirin had no effect on bleeding (Hahn 1979). Despite the well‐documented reduction in IUD‐related bleeding, prophylactic ibuprofen administration had no impact on IUD continuation rates over the first year in the large Chilean trial (Hubacher 2006). Thus, the objective reduction in bleeding may not appreciably influence the subjective acceptability of IUDs. The impact of prophylactic use of NSAIDs on pain is less clear. A recent Danish trial found no benefit of prophylactic ibuprofen at IUD insertion (Jensen 1998), whereas a Finnish trial reported less pain after insertion with prophylactic use of tolfenamic acid than with placebo (Ylikorkala 1978). Naproxen appeared beneficial with insertion of the Dalkon Shield, but the relevance of this observation to contemporary IUDs is limited (Massey 1974). Treatment trials NSAIDs also reduced bleeding in IUD users with heavy blood loss; naproxen (Davies 1981), indomethacin (Yarkoni 1984), suprofen (Di Lieto 1987), and diclofenac sodium (Ylikorkala 1983) all proved superior to placebo. Mefenamic acid was comparable to desmopressin in reducing heavy bleeding (Mercorio 2003), but the greater cost and more frequent side effects argue against use of desmopressin. In contrast, tranexamic acid reduced bleeding more than did either diclofenac sodium or placebo, but it was associated with more unpleasant side effects (Ylikorkala 1983). The superiority of tranexamic acid over NSAIDs is consistent with the conclusion of trials with women not using IUDs (Lethaby 2007). Most trials found NSAIDs to be effective therapy for IUD‐related pain. NSAIDs found superior to placebo included naproxen sodium and suprofen (Buttram 1979; Di Lieto 1987; Lalos 1983). In contrast, one trial found no benefit of diclofenac sodium on dysmenorrhea (Ylikorkala 1983). Evidence is insufficient to recommend a specific NSAID or regimen. Few trials have compared one NSAID to another, and one trial (Toppozada 1982) that compared several NSAIDs found no important differences in effect on bleeding. Similarly, for women with heavy bleeding unrelated to IUD use, naproxen and mefenamic acid appeared equally helpful (Lethaby 2007). Hence, the choice of an NSAID may be based on other concerns, such as cost, convenience, and side effects. For example, naproxen sodium requires twice daily dosing, in contrast to ibuprofen, which is taken four times daily. Another Cochrane review concluded that gastrointestinal side effects may be greater with naproxen than with mefenamic acid (Lethaby 2007). Similarly, for women not using an IUD, the preferred NSAID for treating dysmenorrhea (Marjoribanks 2003) remains unsettled as well. An alternative approach to managing bleeding and pain with IUD use is to switch to a levonorgestrel intrauterine system, which reduces both menstrual bleeding and dysmenorrhea (Lethaby 2005; Pakarinen 2001). No serious harms were reported in these trials, although such reporting was incomplete (Ioannidis 2002), especially in reports published in the 1970s and 1980s. The anticipated gastrointestinal side effects of NSAIDs did not appear problematic in the regimens tested. Trial reports varied widely in quality, but reporting of randomization method and allocation concealment was incomplete for most. Attempts to contact investigators were largely unproductive, due to the age of the reports. Subjective outcomes, lack of blinding, and losses after randomization likely biased the Chinese trial of Baofuxin (Wu 2000). The claim of double blinding appears to be incorrect. Moreover, the report suggested a lingering effect of the traditional medicine for up to 90 days after discontinuation, which appears improbable. If participants assigned to placebo took unauthorized, over‐the‐counter NSAIDs to compensate for their lack of treatment benefit, this would have biased the comparison toward the null. If this occurred, then the true differences between NSAID and placebo may be larger than reported. Authors' conclusions Implications for practice. NSAIDs should be considered as first‐line therapy of bleeding and pain associated with IUD use. The optimal NSAID and regimen are unclear, so the choice of NSAID may be based on cost, convenience, or side effects instead. Should NSAID treatment not yield the desired improvement in bleeding, tranexamic acid, where available, could then be considered. Because of its high cost, desmopressin should not be used. Existing evidence does not support prophylactic ibuprofen administration with the first six menses after IUD insertion. Implications for research. Trials comparing leading NSAIDs with each other might help establish the best approach. However, since few important differences are evident between NSAIDs used as analgesics, comparative trials of effects on bleeding or pain associated with IUD use should not be a research priority. What's new | Date | Event | Description | |---|---|---| | 18 August 2011 | New search has been performed | Searches were updated. No new trials for inclusion were found. Updated one ongoing study Madden 2011 and added another Bayer 2011. | History Protocol first published: Issue 2, 2006 Review first published: Issue 4, 2006 | Date | Event | Description | |---|---|---| | 20 February 2009 | New search has been performed | Searches were updated; searches of clinical trials databases were added. A trial was added to 'ongoing studies' that had not yet begun enrollment. | | 15 April 2008 | Amended | Converted to new review format. | | 30 September 2006 | New citation required and conclusions have changed | Substantive amendment |

Carol Manion of FHI360 coordinated the initial literature searches. Appendices Appendix 1. Search strategies, 2011 MEDLINE via PubMed (2009 to 17 Aug 2011) (intrauterine devices OR intrauterine system* OR IUD* OR IUC* OR IUS) AND (NSAID* OR anti inflammatory agents, nonsteroidal) CENTRAL (2009 to 17 Aug 2011) contracept* in Title, Abstract or Keywords AND (non‐steroidal anti‐inflammatory OR NSAID*) in Title, Abstract or Keywords POPLINE (17 Aug 2011 and past five years) (iud*/iuc*/intrauterine device*) & (NSAID*/nonsteroidal anti inflammatory agent*) LILACS (17 Aug 2011) intrauterine devices or intrauterine device or dispositivos intrauterinos or dispositivos Intra‐uterinos OR IUD or iuds OR IUC or IUCD or iucds [Words] and bleeding or pain [Words] ClinicalTrials.gov (01 Jan 2009 to 18 Aug 2011) Search terms: non‐steroidal anti‐inflammatory OR NSAID* Intervention: contraceptive OR contraception ICTRP Title or Intervention: non‐steroidal anti‐inflammatory OR NSAID Condition: contraception OR contraceptive Appendix 2. Previous search strategies MEDLINE via PubMed (intrauterine devices OR intrauterine system* OR IUD* OR IUCD* OR IUS) AND (NSAID* OR anti inflammatory agents, nonsteroidal) CENTRAL contracept* AND (non‐steroidal anti‐inflammatory OR NSAID*) in Title, Abstract or Keywords POPLINE (iud*/iucd*/intrauterine device*) & (NSAID*/nonsteroidal anti inflammatory agent*) LILACS intrauterine devices or intrauterine device or dispositivos intrauterinos or dispositivos Intra‐uterinos OR IUD or iuds OR IUCD or iucds [Words] and bleeding or pain [Words] EMBASE (INTRAUTERINE CONTRACEPTIVE DEVICE or INTRAUTERINE(1N)DEVICE? or IUD? OR IUCD?) and (NONSTEROID ANTI INFLAMMATORY AGENT or NONSTEROID ANTI INFLAMMATORY DRUG or NONSTEROID ANTIINFLAMMATORY AGENT or NONSTEROIDAL(1N)ANTI(1N)INFLAMMATORY(1N) AGENT or NONSTEROIDAL(1N)ANTI(1N)INFLAMMATORY(1N)DRUG or NONSTEROIDAL(1N)ANTIINFLAMMATORY(1N)AGENT? or NONSTEROIDAL(1N)ANTIINFLAMMATORY(1N)DRUG? or NSAID?) CINAHL (intrauterine device* or IUD* or IUCD*) and (non?steroidal anti?inflammatory agent? or non?steroidal anti?inflammatory drug? or non?steroidal anti?inflammatory agent* or non?steroidal anti?inflammatory drug* or NSAID*) ClinicalTrials.gov Search terms: non‐steroidal anti‐inflammatory OR NSAID Condition: contraceptive OR contraception ICTRP Title or Intervention: non‐steroidal anti‐inflammatory OR NSAID Condition: contraception OR contraceptive Data and analyses Comparison 1. Naproxen 1250 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 47 | Mean Difference (IV, Fixed, 95% CI) | ‐36.0 [‐57.17, ‐14.83] | Comparison 2. Naproxen 1000 mg (Day 1) then 750 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 49 | Mean Difference (IV, Fixed, 95% CI) | ‐40.0 [‐55.40, ‐24.60] | Comparison 3. Naproxen 1250 mg versus naproxen 1000 mg (Day 1) then 750 mg per day. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 40 | Mean Difference (IV, Fixed, 95% CI) | 4.0 [‐16.11, 24.11] | Comparison 4. Suprofen 800 mg (Day 1) then 600 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Moderate or "intense" decrease in bleeding | 1 | 56 | Odds Ratio (M‐H, Fixed, 95% CI) | 197.31 [10.55, 3691.66] | | 2 Moderate or "intense" relief of pain | 1 | 56 | Odds Ratio (M‐H, Fixed, 95% CI) | 124.2 [13.52, 1141.14] | Comparison 5. Aspirin 1500 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 20 | Mean Difference (IV, Fixed, 95% CI) | ‐1.5 [‐29.60, 26.60] | Comparison 6. Ibuprofen 600 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Median pain values | Other data | No numeric data | 6.1. Analysis. Comparison 6 Ibuprofen 600 mg per day versus placebo, Outcome 1 Median pain values. | Median pain values | ||| |---|---|---|---| | Study | Time period | Ibuprofen | Placebo | | Jensen 1998 | at insertion | 3.3 | 2.5 | | Jensen 1998 | first 4 to 6 hours | 1.7 | 1.8 | | Jensen 1998 | first day after insertion | 1.4 | 1.3 | | Jensen 1998 | second day after insertion | 1.3 | 1.1 | | Jensen 1998 | third day after insertion | 1.1 | 1.1 | | Jensen 1998 | Comparison 7. Aspirin 1500 mg per day versus paracetamol 1500 mg per day. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 20 | Mean Difference (IV, Fixed, 95% CI) | ‐0.30 [‐26.16, 25.56] | Comparison 8. Mefenamic acid 1500 mg per day versus desmopressin 300 mcg nasally per day. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Pictorial blood loss assessment chart | 1 | 24 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [0.0, 0.0] | Comparison 9. Iboprofen 1600 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 40 | Mean Difference (IV, Fixed, 95% CI) | ‐14.11 [‐36.04, 7.82] | | 1.1 Copper IUD users | 1 | 20 | Mean Difference (IV, Fixed, 95% CI) | ‐9.47 [‐34.46, 15.52] | | 1.2 Lippes Loop users | 1 | 20 | Mean Difference (IV, Fixed, 95% CI) | ‐29.68 [‐75.44, 16.08] | Comparison 10. Indomethacin 100 mg daily or flufenamic acid 600 mg daily or alclofenac 1500 mg daily versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 36 | Mean Difference (IV, Fixed, 95% CI) | ‐39.8 [‐42.84, ‐36.76] | | 2 Severe or moderate pain | 1 | 36 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.00 [0.00, 0.06] | Comparison 11. Indomethacin 50 mg per day versus Baofuxin 16 g per day. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Bleeding/spotting days | 1 | 144 | Mean Difference (IV, Fixed, 95% CI) | 0.5 [‐2.96, 3.96] | | 2 Bleeding/spotting episodes | 1 | 144 | Mean Difference (IV, Fixed, 95% CI) | 0.10 [‐0.38, 0.58] | Comparison 12. Indomethacin 100 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 36 | Mean Difference (IV, Fixed, 95% CI) | ‐85.22 [‐119.27, ‐51.17] | Comparison 13. Diclofenac sodium 150 mg then 75 mg versus tranexamic acid 4.5 mg per day. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 38 | Mean Difference (IV, Fixed, 95% CI) | 42.70 [12.05, 73.35] | Comparison 14. Diclofenac sodium 150 mg (Day 1) then 75 mg on days 2‐5 versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Menstrual blood loss | 1 | 38 | Mean Difference (IV, Fixed, 95% CI) | ‐26.20 [‐66.77, 14.37] | Comparison 15. Tolfenamic acid 600 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Duration of bleeding after IUD insertion | 1 | 122 | Mean Difference (IV, Fixed, 95% CI) | ‐0.60 [‐1.51, 0.31] | | 2 Number of sanitary pads used for bleeding immediately after IUD insertion | 1 | 122 | Mean Difference (IV, Fixed, 95% CI) | ‐1.4 [‐2.06, ‐0.74] | | 3 Next menses more abundant than normal | 1 | 310 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.54 [0.34, 0.85] | | 4 Clots in menstrual blood with subsequent menses | 1 | 310 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.43 [0.27, 0.68] | | 5 Subsequent menses more painful than normal | 1 | 310 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.71 [0.44, 1.14] | Comparison 16. Naproxen sodium 550 mg then 275 mg every 6 hr as needed versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 "Good" or "excellent" relief of uterine pain in one or more cycles | 1 | 33 | Odds Ratio (M‐H, Fixed, 95% CI) | 14.0 [2.60, 75.41] | | 2 Mean daily pain relief scores | 1 | 27 | Mean Difference (IV, Fixed, 95% CI) | 3.10 [0.27, 5.93] | Comparison 17. Naproxen 500 mg initially then additional doses of 250 mg up to 1250 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Pain relief rated better than alternative treatment | 1 | 40 | Odds Ratio (M‐H, Fixed, 95% CI) | 32.11 [5.66, 182.18] | Comparison 18. Naproxen 1200 mg per day versus placebo. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 Need for additional analgesia after IUD insertion | 1 | 50 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.22 [0.07, 0.72] | Comparison 19. Ibuprofen 1200 mg per day versus placebo with 6 menses after IUD insertion. | Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size | |---|---|---|---|---| | 1 IUD removal by 26 weeks for dysmenorrhea or increased menstrual blood loss | 1 | 1962 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.96 [0.66, 1.38] | | 2 IUD removal by 52 weeks | 1 | 1962 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.05 [0.86, 1.28] | Characteristics of studies Characteristics of included studies [ordered by study ID] Buttram 1979. | Methods | Randomized controlled trial | | | Participants | 35 women in the U.S. with either uterine pain starting after IUD insertion or prior menstrual pain exacerbated by IUD insertion. Inclusion criteria included a normal physical examination, and, if necessary for pain evaluation, normal laparoscopy. Women with cycle irregularities were excluded. | | | Interventions | Naproxen sodium 550 mg initially, then 275 mg every six hours as needed for uterine pain versus a placebo (lactose). Trial ran for three cycles. | | | Outcomes | Uterine pain as judged by six‐point Likert scale. | | | Notes | Method of randomization not specified; no mention of allocation concealment. Reportedly double blind; placebos used. Sample size calculation not provided. Thirty‐five women initially enrolled; two dropped out for personal reasons. Only 27 women used the assigned drugs for each of three menstrual cycles. Of the six who did not complete all planned treatments, three did not have satisfactory relief, two had no further pain, and one had her IUD removed. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Unclear risk | No mention | Davies 1981. | Methods | Randomized controlled trial with crossover | | | Participants | 48 women of reproductive age using IUDs in England with measured menstrual blood loss averaging at least 80 mL per cycle over two pre‐treatment cycles. Participants had regular cycles, were healthy, and had a normal pelvic examination. | | | Interventions | Three treatment arms: naproxen 1250 mg per day, naproxen 1000 mg on first day then 750 mg on subsequent days, or placebo. Each participant took two of the three treatments, each for two menstrual cycles. Each regimen was begun on the first day of menstrual bleeding and continued for a total of five days. | | | Outcomes | Volume of menstrual blood loss measured by alkaline hematin method; menstrual diaries kept by participants. | | | Notes | Complex design with small number of participants. Twenty‐nine percent of participants were lost or excluded after randomization. Method of randomization not stated, and no mention of allocation concealment. No sample size calculation provided. Placebos used to blind participants and investigators, but whether this worked is unclear. One participant withdrew from the trial because of nausea and vomiting considered related to treatment. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Unclear risk | No mention | Di Lieto 1987. | Methods | Probably randomized controlled trial with crossover | | | Participants | 28 women, aged 21 to 33 years, in Italy using IUDs and who "suffered from increased menstrual blood loss and pelvic pain." | | | Interventions | Suprofen 800 mg per day on day one, then 600 mg per day versus placebo. Each regimen was started at the first sign of bleeding and continued up to seven days maximum. Each participant had one pre‐treatment month, one month with suprofen, then the next month with the placebo. | | | Outcomes | Likert‐like scale of subjective improvement of bleeding and pain. | | | Notes | Method of randomization not specified; no mention of allocation concealment. Placebo used to enable blinding of participants and investigators; report does not indicate whether blinding was successful. No sample size calculation. No losses after randomization. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Unclear risk | No mention | Hahn 1979. | Methods | Randomized controlled trial with crossover | | | Participants | 37 women aged 20 to 40 years in Sweden, 10 of whom were using copper IUDs | | | Interventions | Aspirin 1500 mg per day versus paracetamol 1500 mg per day versus placebo. Each participant received each treatment arm for one month. | | | Outcomes | Menstrual blood loss measured by alkaline hematin method | | | Notes | Method of randomization not specified; authors reported using "double blind crossover technique." Communication with author indicated allocation concealment by "sealed envelopes." Sample size was one of convenience. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Low risk | Communication with author indicated allocation concealment by "sealed envelopes." | Hubacher 2006. | Methods | Randomized controlled trial | | | Participants | 2019 women aged 18‐49 yr. in Chile having a copper T380A IUD inserted. Inclusion criteria included literacy, no previous IUD use, more than six weeks postpartum if recently pregnant, and without contraindications to an IUD or ibuprofen. | | | Interventions | Ibuprofen 1200 mg per day versus placebo, to be taken with the first six menses after IUD insertion and for a maximum of 5 days at a time. | | | Outcomes | The primary outcome was IUD removal within 12 months of insertion. Participants were followed up at 6, 13, 26, and 52 weeks after insertion. | | | Notes | Randomization by random permuted blocks with block sizes of 20, 10, 4, and 2, with computer generation of sequence. Allocation concealment by pharmacy distribution of medicine bottles. Sample size was calculated to achieve 85% power to detect a reduction in discontinuation from 16% to 11% with an alpha of 0.025 (one‐sided hypothesis) Fifty‐seven women did not have an IUD inserted; these were excluded from the primary analysis. Thirty‐five minor protocol violations occurred, but these women were analyzed within the groups to which they were initially assigned, according to intention‐to‐treat principles. 95% of participants provided 52‐week follow‐up information. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Low risk | Pharmacy distribution of medicine bottles | Jensen 1998. | Methods | Randomized controlled trial with four treatment groups | | | Participants | 55 women in Denmark having insertion of either a Nova T or Gyne T 380 Slimline IUD inserted. Trial conducted from May 1994 to May 1995. Trial enrolled all Danish‐speaking women in good health who requested an IUD in the investigators' clinic. Exclusion criteria included age <18 years, serious illness, allergy to NSAID or aspirin, and medication of any kind except for oral contraceptives. | | | Interventions | Ibuprofen 1800 mg in divided doses (before insertion, four to six hours after insertion, and next morning) versus placebo. The type of IUD to be inserted was randomly chosen. | | | Outcomes | Pain assessed by a 10‐point scale, administered during insertion, four to six hours after insertion, and again during the first three days. | | | Notes | Computer‐generated randomization. Correspondence with author indicated that allocation concealment was achieved by central pharmacy packaging of drug in sealed envelopes. Randomization stratified by type of IUD inserted. Neither participants nor investigators were aware of treatment assignment. Tablets were of same size and shape, but taste may have differed. Formal a priori sample size calculation. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Low risk | Correspondence with author indicated that allocation concealment was achieved by central pharmacy packaging of drug in sealed envelopes. | Lalos 1983. | Methods | Randomized controlled trial with crossover. | | | Participants | 21 women in Sweden with primary dysmenorrhea aggravated by insertion of a copper IUD (either a T or the Cu‐7). Inclusion criteria included severe dysmenorrhea, regular menstrual cycles, and normal gynecological examination. | | | Interventions | Naproxen 500 mg followed by 250 mg two to four times a day, with a maximum of 1250 mg, versus placebo. The drug was started at the first sign of menstrual distress. | | | Outcomes | Pain assessed by a five‐point Likert scale. | | | Notes | Communication with the investigators indicated that a table of random numbers was used to generate the random sequence. Allocation concealment was not described. Participants and investigators were blinded as to treatment by using identical‐appearing placebo tablets that had the same taste as the naproxen; both the drug and the placebo were distributed in amber bottles. Sample‐size calculation not provided. One participant excluded after randomization because of intervening operation. Informed consent was verbal, not written. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Low risk | Identical‐appearing placebo tablets had the same taste as the naproxen. Both the drug and the placebo were distributed in amber bottles. | Massey 1974. | Methods | Randomized controlled trial with blinding | | | Participants | 50 women in the US having a Dalkon Shield inserted. Inclusion criteria included normal physical examination and laboratory tests. Exclusion criteria included severe painful, menses or premenstrual tension. 48 women were nulliparous, and two were primiparous. | | | Interventions | Naproxen 300 mg taken the evening before IUD insertion, 300 mg 1.5 hours before insertion, and 300 mg at two and four hours after insertion versus placebo. | | | Outcomes | Requirement for additional analgesia plus complex pain scores created for the trial. | | | Notes | Random permuted blocks with a block size of 10; source of the random sequence not specified. Allocation concealment not mentioned. Sample size not explained. Two women assigned to naproxen did not have IUD inserted and were replaced by another randomization, resulting in 24 assigned to naproxen and 26 to placebo. Blinding accomplished by providing four identical capsules in strips. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Unclear risk | No mention | Mercorio 2003. | Methods | Randomized controlled trial without blinding | | | Participants | 24 women aged 20 to 40 years with IUD‐associated menorrhagia in Italy. Inclusion criteria included pictorial blood loss assessment for two treatment cycles indicating menorrhagia (score of 100 or greater); exclusion criteria included gynecological pathology, family or personal history of bleeding disorders, or prior use of study drugs. | | | Interventions | Desmopressin 300 mcg intranasal spray each morning for the first five days after start of menses versus mefenamic acid 1500 mg per day for the first 5 days after the start of menses. Study duration of three months. | | | Outcomes | Pictorial blood loss assessment charts | | | Notes | Randomization done by "computer analysis" and allocation concealment by "envelope." Sample size calculation not provided. No participants discontinued the trial prematurely. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Low risk | By "envelope" | Roy 1981. | Methods | Randomized controlled crossover trial with blinding of treatment for participants and investigators | | | Participants | 20 women in U.S.A. aged 18 to 33 years without bleeding or other complaints; ten used copper IUDs and ten used Lippes Loop. Inclusion criteria included having used the IUD for at least five months and being free of side effects. Exclusion criteria included previous salpingitis, undiagnosed bleeding, uterine anomalies, multiple fibroids, and contraindications to ibuprofen. | | | Interventions | Ibuprofen 1600 mg per day starting with bleeding and continuing until end of bleeding or maximum of seven days. Each participant had one month of observation, one month on ibuprofen or placebo, and one month crossed over to the alternative treatment. | | | Outcomes | Menstrual blood loss measured by alkaline hematin method | | | Notes | Communication with investigator indicated computer‐generated randomization sequence. Allocation concealment by pharmacy distribution of active drug and identical‐appearing placebo. Sample size was based on feasibility rather than power calculation. Randomization was designed "to ensure equal and random patient distribution" to the two groups. Analysis was stratified by copper IUD versus Lippes Loop. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Low risk | Pharmacy distribution of active drug and identical‐appearing placebo | Toppozada 1982. | Methods | Randomized controlled crossover trial without blinding | | | Participants | 18 parous women in Egypt aged 20 to 35 years using a copper IUD. Inclusion criteria included regular menstruation and hemoglobin > 12 g%. Exclusion criteria included peptic ulcer disease, salpingitis, or "allergic" disease. | | | Interventions | Indomethacin 100 mg per day versus flufenamic acid 600 mg per day versus alclofenac 1500 mg per day. Treatment started on the first day of bleeding and was given for three days. Observation period ran four months. In each treatment arm, half started on active drug and half on placebo; after two months, crossover to the alternative occurred. Thus, each participant had two months on the assigned NSAID and two months on placebo. | | | Outcomes | Menstrual blood loss measured by alkaline hematin method | | | Notes | Communication with investigator indicated computer‐generated randomization scheme; allocation concealment by sequentially‐numbered, sealed, opaque envelopes provided by WHO. Sample size calculation not provided. No apparent losses to follow up after randomization. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Low risk | Sequentially‐numbered, sealed, opaque envelopes provided by WHO | Wu 2000. | Methods | Randomized controlled trial without blinding | | | Participants | 180 parous women in China. All had complaints of heavy or prolonged menses (more than seven days), good general health, a copper IUD for at least six months. Exclusion criteria included hemorrhagic illness or leiomyomas, unwillingness to avoid other medicines, and unwillingness to maintain a menstrual diary. | | | Interventions | The NSAID regimen was indomethacin 50 mg daily in divided doses for 7 days, starting on the first day of menses. The alternative regimen was Baofuxin (a traditional Chinese medicine containing Tangshen, Radix Astragali seu Hedysari, Ass‐hide Asafetida, Hairyvein Agrimonia Herb, Indian Madder Root and other ingredients in a granular compound). The 8 g bag was dissolved in hot water and drunk twice daily for 10 days, starting on the first day of menses. | | | Outcomes | Menstrual patterns as determined by diaries. Menstrual diaries were maintained for 90 days before, 3 menstrual cycles of treatment, and 90 days after stopping treatment. | | | Notes | Method of randomization not specified; no mention of allocation concealment. Sample size calculation not provided. Lack of blinding likely influenced acceptability of traditional Chinese medicine. Report states "double‐blind" trial, but no description of double dummy or other means to disguise treatment. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Unclear risk | No mention | Yarkoni 1984. | Methods | Randomized controlled trial with blinding and crossover | | | Participants | 15 Israeli women aged 24 to 36 years with Lippes Loop C who had mean measured blood loss of >100 mL by alkaline hematin method. Exclusion criteria included history of pelvic inflammatory disease, bleeding disorders, or leiomyomas. Those with contraindications to NSAIDs were excluded as well. | | | Interventions | Indomethacin 100 mg daily for four days, starting on the first day of bleeding versus placebo. After two months of baseline measurements in 35 women, the 15 with mean losses > 100 mL were randomized. Each participant randomly received either indomethacin or placebo each month; each participant received two months of each regimen, not necessarily in sequence. | | | Outcomes | Menstrual blood loss as measured by the alkaline hematin method | | | Notes | Method of randomization not specified; allocation concealment not described. Sample size calculation not provided. High discontinuation proportion; only 9 of 15 women enrolled completed the study. Standard deviations not provided for mean blood loss. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Unclear risk | No mention | Ylikorkala 1978. | Methods | Randomized controlled trial with blinding of participants and investigators | | | Participants | 160 Finnish women of unspecified ages requesting insertion of a copper T200 IUD. Inclusion criteria included normal history and physical examination. | | | Interventions | Tolfenamic acid 600 mg per day starting at IUD insertion and continuing for seven days; this was to be repeated during the next three menses. Placebo given in same regimen. | | | Outcomes | Subjective reports of pain and bleeding patterns | | | Notes | Method of randomization not specified; no mention of allocation concealment. Sample size calculation not provided. Blinding of participants and investigators by using coded pill bottles with either tolfenamic acid of placebo; report does not state that placebo was identical in appearance and taste. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Unclear risk | No mention | Ylikorkala 1983. | Methods | Randomized controlled trial with blinding and crossover | | | Participants | 19 Finnish women aged 25 to 44 years with measured mean menstrual blood loss >70 mL. 18 were using a copper IUD and one a non‐medicated device. All had normal pelvic and general physical examinations. | | | Interventions | Tranexamic acid 4.5 g per day for five days versus diclofenac sodium 150 mg on day one then 75 mg per day on days two through four versus placebo. Over the five months of the trial, each participant received placebo for one month, two months of tranexamic acid, and two months of diclofenac sodium. | | | Outcomes | Menstrual blood loss | | | Notes | Method of randomization not specified; allocation concealment not described. No sample size calculation provided. Participants and investigators kept blinded as to treatment by use of double dummy. | | | Risk of bias | || | Bias | Authors' judgement | Support for judgement | | Allocation concealment (selection bias) | Unclear risk | No mention | Characteristics of excluded studies [ordered by study ID] | Study | Reason for exclusion | |---|---| | Blum 1984 | Not a randomized controlled trial. | | Davies 1980 | Not a randomized controlled trial. | | Dreher 1980 | Not a randomized controlled trial. | | Guillebaud 1978 | Before‐after study. | | Makarainen 1986 | Participants allocated to treatment by odd‐even year of birth. | | Pedron 1982 | Before‐after study. | | Pedron 1983 | Before‐after study. | | Pizarro 1988 | Correspondence with author revealed alternate assignment to treatment groups. | | Sahmay 1987 | Random allocation not specified in methods. Allocation of 43 participants to 79 cycles of proquazon and 50 cycles of placebo is unlikely to have occurred by simple randomization. | | Toppozada 1980 | Methods did not specify random allocation to treatments. | Characteristics of ongoing studies [ordered by study ID] Bayer 2011. | Trial name or title | Management of initial bleeding/spotting associated with the levonorgestrel‐releasing intrauterine system (MIRENA) | | Methods | RCT | | Participants | 186 women, 18 to 45 years old. Inclusion criteria: healthy female requesting contraception, successful interval insertion of MIRENA, regular menstrual periods, normal or clinically insignificant cervical smear. Exclusion criteria: pregnancy or lactation; climacteric symptoms prior to screening; known or suspected clinically significant ovarian cysts, endometrial polyps, fibroids, or other genital organ pathology; undiagnosed abnormal genital bleeding; history of thromboembolic disease or risk factors for venous thromboembolism; current migraine, focal migraine with asymmetrical visual loss or other symptoms indicating transient cerebral ischemia, or exceptionally severe headaches; hypersensitivity to ingredient of investigational products or non‐investigational product; daily or frequent use of NSAIDs for any condition; not willing to use NSAIDs as pain medication | | Interventions | Levonorgestrel IUS with either mefenamic acid, tranexamic acid or placebo; 90‐day treatment period | | Outcomes | Primary: cumulative number of bleeding or spotting days Secondary: bleeding patterns observed in women during treatment period; bleeding patterns observed in women during follow‐up period; satisfaction with oral treatment for bleeding or spotting; dysmenorrhea; continuation rate with study drug; continuation rate with Mirena; adverse events | | Starting date | Mar 2011; estimated completion Dec 2011 | | Contact information | clinical‐trials‐[email protected] | | Notes | Currently recruiting | Madden 2011. | Trial name or title | A randomized controlled trial of oral naproxen and transdermal estradiol for bleeding in LNG‐IUC | | Methods | RCT | | Participants | 129 women from 18 to 45 years old Inclusion Criteria: choosing LNG‐IUC for contraception; English‐speaking; willing to avoid additional exogenous hormones such as OCs; willing to avoid additional NSAIDS. Exclusion Criteria: known or suspected pregnancy, contraindication to estrogen use, recent or planned major surgery during study, history of severe hepatic disease or liver tumor, known or suspected estrogen‐dependent neoplasm, vaginal bleeding of unknown etiology, cigarette smoking and age over 35 years, contraindication to NSAID use; use of medications that alter estrogen metabolism, i.e. rifampin, certain anti‐seizure medications; regular use of NSAID; menorrhagia, metrorrhagia, symptomatic uterine fibroids, or endometrial polyp; hypersensitivity or allergy to estradiol patch; injectable contraception use within 6 months of study start; delivery or abortion in past 4 weeks; prior use of LNG‐IUD; condition that would contraindicate study participation. | | Interventions | Oral naproxen or transdermal estradiol or placebo for first 12 weeks of use | | Outcomes | Primary: number of bleeding and spotting days Secondary: bleeding patterns; satisfaction with LNG‐IUC; continuation rates of LNG‐IUC | | Starting date | Nov 2008; completed Jan 2011 | | Contact information | TE Madden: [email protected] | | Notes | No study report found; wrote to author about whether a study report was available or in progress | Contributions of authors Dr Grimes developed the idea, registered the topic, and drafted the protocol and review. Dr Hubacher edited the protocol and assisted with the literature search and with writing and editing the review. Dr Lopez edited the protocol, performed an independent data extraction, and helped write the review. For the 2009 and 2011 updates, she conducted the searches and updated the review as needed.. Dr Schulz edited the review and provided statistical oversight. Sources of support Internal sources No sources of support supplied External sources U. S. Agency for International Development, USA. National Institute of Child Health and Human Development, USA. Declarations of interest Dr Grimes has consulted with the pharmaceutical companies Bayer Healthcare Pharmaceuticals and Merck & Co, Inc. Dr Hubacher was the Principal Investigator for one of the trials included. New search for studies and content updated (no change to conclusions)

References to studies included in this review Buttram 1979 {published data only} - Buttram V, Izu A, Henzl MR. Naproxen sodium in uterine pain following intrauterine contraceptive device insertion. American Journal of Obstetrics and Gynecology 1979;134:575‐8. [DOI] [PubMed] [Google Scholar] Davies 1981 {published data only} - Davies AJ, Anderson ABM, Turnbull AC. Reduction by naproxen of excessive menstrual bleeding in women using intrauterine devices. Obstetrics and Gynecology 1981;57:74‐8. [PubMed] [Google Scholar] Di Lieto 1987 {published data only} - Lieto A, Catalano D, Miranda L, Paladini A. Action of a prostaglandin synthetase inhibitor on IUD associated uterine bleeding. Clinical and Experimental Obstetrics and Gynecology 1987;14:41‐4. [PubMed] [Google Scholar] Hahn 1979 {published and unpublished data} - Hahn L, Petruson B. The influence of acetylsalicylic acid and paracetamol on menstrual blood loss in women with and without an intrauterine contraceptive device. American Journal of Obstetrics and Gynecology 1979;135:393‐6. [DOI] [PubMed] [Google Scholar] Hubacher 2006 {published data only} - Hubacher D, Reyes V, Lillo S, Pierre‐Louis B, Zepeda A, Chen P‐L, et al. Preventing copper IUD removals due to side‐effects among first time users: placebo‐controlled randomized trial to study the effect of prophylactic ibuprofen. Human Reproduction 2006;21:1467‐72. [DOI] [PubMed] [Google Scholar] Jensen 1998 {published and unpublished data} - Jensen HH, Blaabjerg J, Lyndrup J. Prophylactic use of prostaglandin synthetase inhibitors in connection with IUD insertion [Profylaktisk brug af prostaglandinsyntesehaemmere ved oplaegning af spiral]. Ugeskrift for Laeger 1998;160:6958‐61. [PubMed] [Google Scholar] Lalos 1983 {published and unpublished data} - Lalos O, Nilsson B. Dysmenorrhea in women with intrauterine contraceptive device. Treatment with a prostaglandin synthetase inhibitor, naproxen. International Journal of Gynaecology and Obstetrics 1983;21:33‐7. [DOI] [PubMed] [Google Scholar] Massey 1974 {published data only} - Massey SE, Varady JC, Henzl MR. Pain relief with naproxen following insertion of an intrauterine device. Journal of Reproductive Medicine 1974;13:226‐31. [PubMed] [Google Scholar] Mercorio 2003 {published data only} - Mercorio F, Simone R, Carlo C, Bifulco G, Tessitore G, Spiezio Sardo A, et al. Effectiveness and mechanism of action of desmopressin in the treatment of copper intrauterine device‐related menorrhagia: a pilot study. Human Reproduction 2003;18:2319‐22. [DOI] [PubMed] [Google Scholar] Roy 1981 {published and unpublished data} - Roy S, Shaw ST Jr. Role of prostaglandins in IUD‐associated uterine bleeding ‐‐ effect of a prostaglandin synthetase inhibitor (ibuprofen). Obstetrics and Gynecology 1981;58:101‐6. [PubMed] [Google Scholar] Toppozada 1982 {published and unpublished data} - Toppozada M, Anwar M, Abdel Rahman H, Gaweesh S. Control of IUD‐induced bleeding by three non‐steroidal anti‐inflammatory drugs. Contraceptive Delivery Systems 1982;3:117‐25. [PubMed] [Google Scholar] Wu 2000 {published data only} - Wu S, Wang C, Cheng W, Han X, Wang S, Qi J, et al. Randomized multi‐center study of baofuxin for treatment of bleeding side‐effect induced by IUD. Reproduction and Contraception 2000;11:152‐7. [Google Scholar] Yarkoni 1984 {published data only} - Yarkoni S, Anteby SO. Treatment of IUD related menorrhagia by indomethacin. Clinical and Experimental Obstetrics and Gynecology 1984;11:120‐2. [PubMed] [Google Scholar] Ylikorkala 1978 {published data only} - Ylikorkala O, Kauppila A, Siljander M. Anti‐prostaglandin therapy in prevention of side effects of intrauterine devices. Lancet 1978;2:393‐5. [DOI] [PubMed] [Google Scholar] Ylikorkala 1983 {published data only} - Ylikorkala O, Viinikka L. Comparison between antifibrinolytic and antiprostaglandin treatment in the reduction of increased menstrual blood loss in women with intrauterine contraceptive devices. British Journal of Obstetrics and Gynaecology 1983;90:78‐83. [DOI] [PubMed] [Google Scholar]

to studies excluded from this review Blum 1984 {published data only} - Blum M, Pery J. Prevention of iron deficiency anemia in IUD users by prostaglandin synthetase inhibitors. Contraceptive Delivery Systems 1984;5:21‐3. [PubMed] [Google Scholar] Davies 1980 {published data only} - Davies AJ. Reduction of menstrual blood loss by naproxen in intrauterine contraceptive device users. International Journal of Gynaecology and Obstetrics 1980;17:353‐4. [DOI] [PubMed] [Google Scholar] Dreher 1980 {published data only} - Dreher E, Fischer B. Treatment of primary dysmenorrhea and dysmenorrhea because of IUD with PG‐synthetase inhibitors. Advances in Prostaglandin and Thromboxane Research 1980;8:1487‐93. [PubMed] [Google Scholar] Guillebaud 1978 {published data only} - Guillebaud J, Anderson ABM, Turnbull AC. Reduction by mefenamic acid of increased menstrual blood loss associated with intrauterine contraception. British Journal of Obstetrics and Gynaecology 1978;85:53‐62. [DOI] [PubMed] [Google Scholar] Makarainen 1986 {published data only} - Makarainen L, Ylikorkala O. Ibuprofen prevents IUCD‐induced increases in menstrual blood loss. British Journal of Obstetrics and Gynaecology 1986;93:285‐8. [DOI] [PubMed] [Google Scholar] Pedron 1982 {published data only} - Pedron N, Lozano M, Aznar R. Treatment of hypermenorrhea with mefenamic acid in women using IUDs. Contraceptive Delivery Systems 1982;3:135‐9. [PubMed] [Google Scholar] Pedron 1983 {published data only} - Pedron N, Aznar R. Effect of naproxen on IUD‐induced hypermenorrhea. Contraceptive Delivery Systems 1983;4:39‐42. [PubMed] [Google Scholar] Pizarro 1988 {published data only} - Pizarro E, Mehech G, Hidalgo M, Munoz G, Romero C. Effect of meclofenamic acid on menstruation in hypermenorrheic women using intrauterine devices [Efecto del acido meclofenamico sobre la perdida de sangre menstrual en mujeres hipermenorreicas usuarias de dispositivo intrauterino]. Revista Chilena de Obstetricia y Ginecologia 1988;53:43‐56. [PubMed] [Google Scholar] Sahmay 1987 {published data only} - Sahmay S, Ertungealp E, Arvas M, Atasu T. Antiprostaglandins in therapy for side effects of intrauterine contraceptive devices [Rahim ici kontraseptif araclarin yan etkilerinin tedavisinde anti‐prostaglandinler]. Cerrahpasa Tip Fakultesis Dergisi 1987;18:75‐83. [PubMed] [Google Scholar] Toppozada 1980 {published data only} - Toppozada M, El‐Attar A, El‐Ayyat MA, Khamis Y. Management of uterine bleeding by PGs or their synthesis inhibitors. Advances in Prostaglandin and Thromboxane Research 1980;8:1459‐63. [PubMed] [Google Scholar]

to ongoing studies Bayer 2011 {published data only} - Bayer HealthCare AG. Management of initial bleeding/spotting associated with the levonorgestrel‐releasing intrauterine system (MIRENA). http://clinicaltrials.gov/ct2/show/NCT01295294 (accessed 17 Aug 2011). [NCT01295294] Madden 2011 {published data only} - Madden TE. A randomized controlled trial of oral naproxen and transdermal estradiol for bleeding in LNG‐IUC. http://clinicaltrials.gov/ct2/show/NCT00789802 (accessed 17 Aug 2011). [NCT00789802] Additional references Allen 2009 - Allen RH, Bartz D, Grimes DA, Hubacher D, O’Brien P. Interventions for pain with intrauterine device insertion. Cochrane Database of Systematic Reviews 2009, Issue 3. [DOI: 10.1002/14651858.CD007373.pub2] [DOI] [PubMed] [Google Scholar] Connell 2002 - Connell EB. The contraception sourcebook. Chicago: Contemporary Books, 2002. [Google Scholar] Elbourne 2002 - Elbourne DR, Altman DG, Higgins JPT, Curtin F, Worthington HV, Vail A. Meta‐analyses involving cross‐over trials: methodological issues. International Journal of Epidemiology 2002;31:140‐9. [DOI] [PubMed] [Google Scholar] Gallegost 1978 - Gallegost AJ, Aznar R, Merino G, Guizer E. Intrauterine devices and menstrual blood loss. A comparative study of eight devices during the first six months of use. Contraception 1978;17:153‐61. [DOI] [PubMed] [Google Scholar] Guillebaud 1976 - Guillebaud J, Bonnar J, Morehead J, Matthews A. Menstrual blood‐loss with intrauterine devices. Lancet 1976;1(7956):387‐90. [DOI] [PubMed] [Google Scholar] Hallberg 1964 - Hallberg L, Nilsson L. Determination of menstrual blood loss. Scandinavian Journal of Laboratory Investigation 1964;16:244‐8. [PubMed] [Google Scholar] Higgins 2005 - Higgins JPT, Green S, editors. Cochrane handbook for systematic reviews of interventions 4.2.5 [updated May 2005]. www.cochrane.org/resources.handbook.hbook.htm. Chichester, UK: John Wiley & Sons, Ltd, (accessed 30 December 2005). Higham 1990 - Higham JM, O'Brien PM, Shaw RW. Assessment of menstrual blood loss using a pictorial chart. British Journal of Obstetrics and Gynaecology 1990;97:734‐9. [DOI] [PubMed] [Google Scholar] Ioannidis 2002 - Ioannidis JP, Chew P, Lau J. Standardized retrieval of side effects data for meta‐analysis of safety outcomes. A feasibility study in acute sinusitis. Journal of Clinical Epidemiology 2002;55:619‐26. [DOI] [PubMed] [Google Scholar] Israel 1974 - Israel R, Shaw ST Jr, Martin MA. Comparative quantitation of menstrual blood loss with the Lippes Loop, Dalkon Shield, and copper T intrauterine devices. Contraception 1974;10:63‐71. [DOI] [PubMed] [Google Scholar] Lethaby 2005 - Lethaby AE, Cooke I, Rees M. Progesterone or progestogen‐releasing intrauterine systems for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2005, Issue 4. [DOI: 10.1002/14651858.CD002126] [DOI] [PubMed] [Google Scholar] Lethaby 2007 - Lethaby A, Augood C, Duckitt K, Farquhar C. Nonsteroidal anti‐inflammatory drugs for heavy menstrual bleeding. Cochrane Database of Systematic Reviews 2007, Issue 4. [DOI: 10.1002/14651858.CD000400] [DOI] [PubMed] [Google Scholar] Marjoribanks 2003 - Marjoribanks J, Proctor ML, Farquhar C. Nonsteroidal anti‐inflammatory drugs for primary dysmenorrhoea. Cochrane Database of Systematic Reviews 2005, Issue 4. [DOI: 10.1002/14651858.CD001751] [DOI] [PubMed] [Google Scholar] Mayo Clinic 2005 - Mayo Clinic. NSAIDs: how to avoid side effects. http://www.mayoclinic.com/health/pain‐medications/PN00058 (accessed 28 December 2005). Milsom 1995 - Milsom I, Andersson K, Jonasson K, Lindstedt G, Rybo G. The influence of the Gyne‐T 380S IUD on menstrual blood loss and iron status. Contraception 1995;52:175‐9. [DOI] [PubMed] [Google Scholar] Pakarinen 2001 - Pakarinen P, Toivonen J, Luukkainen T. Therapeutic use of the LNG IUS, and counseling. Seminars in Reproductive Medicine 2001;19:365‐72. [DOI] [PubMed] [Google Scholar] Reid 2000 - Reid PC, Coker A, Coltart R. Assessment of menstrual blood loss using a pictorial chart: a validation study. British Journal of Obstetrics and Gynaecology 2000;107:320‐2. [DOI] [PubMed] [Google Scholar] Speroff 1996 - Speroff L, Darney PD. A clinical guide for contraception. Second Edition. Baltimore, MD: Williams & Wilkins, 1996. [Google Scholar] United Nations 2003 - United Nations. World Contraceptive Use 2003. http://www.un.org/esa/population/publications/contraceptive2003/WallChart_CP2003.pdf (accessed 27 December 2005). WHO 1997 - World Health Organization. Long‐term reversible contraception. Twelve years of experience with the TCu380A and TCu220C. Contraception 1997;56:341‐52. [PubMed] [Google Scholar] Ylikorkala 1994 - Ylikorkala O. Prostaglandin synthesis inhibitors in menorrhagia, intrauterine contraceptive device‐induced side effects and endometriosis. Pharmacology and Toxicology 1994;75:86‐8. [DOI] [PubMed] [Google Scholar] Zhang 1992 - Zhang JY, Luo LL. Intrauterine device‐induced menorrhagia and endometrial content of prostacyclins. Zhonghua Fu Chan Ke Za Zhi 1992;27:167‐8. [PubMed] [Google Scholar] Zheng 1988 - Zheng XL, Lo LL. Studies on IUD‐induced menorrhagia and increased synthesis of prostacyclin in endometrium. Journal of Tongji Medical University 1988;8:232‐4. [DOI] [PubMed] [Google Scholar]