Description of studies

This review included 33 randomised (n=32) or systematically allocated (n=1) controlled trials (RCTs), conducted in 17 countries (China, Indonesia, Iran, Pakistan, the Philippines and Thailand in Asia; Tunisia and Egypt in Africa; France, the Netherlands, Sweden, Switzerland and the United Kingdom in Europe; the United States of America in North America; Chile and the Dominican Republic in Latin America; and Australia). Studies were conducted at family planning clinics, hospitals and university research centres.

Contraceptives used:
Twelve of the included studies enrolled women using Norplant as their contraceptive method. New Norplant users, as well as women using the method for one to 43 months were enrolled; most of the studies limited enrolment to women using the method for less than one year. Twelve studies were conducted to evaluate interventions in women using the injectable DMPA. Two of these enrolled new users, with the aim of evaluating interventions to prevent bleeding irregularities; the remaining studies enrolled women who had been using the method for one to 18 months. In three trials, women used progestin‐only pills; the duration of use was at least one month in one study and was not defined in the other two studies. Three trials recruited women who had been using Implanon for a period of more than two months. Two studies enrolled new users of the Mirena LNG‐IUS. One study was conducted to evaluate treatment of irregular bleeding in Jadelle users.

Characteristics of participants:

The age of women included in these trials generally was in the range of 20 to 30 years; however, many of the studies' inclusion criteria allowed for the enrolment of women up to age 45.. Two studies (Subakir 2000, and Weisberg 2006) included women 18 to 40 years old. One study (Harel 2002) specifically enrolled adolescent girls; the mean age of participants in this trial was 16 ± 1 years. The parity of women recruited in the trials was reported in only 13 trials and ranged between zero and five. The body weight of study participants was reported in 11 trials and mean body weights were reported in the range between 44 and 73 kg. Height was not mentioned in the majority of the trials, but nine studies reported the body mass index; in these trials BMI ranged from 21 to > 30 kg/m2.

Interventions:

There were 18 types of interventions tested in the trials. Fourteen interventions were tested as therapeutic treatments. These included estrogens alone in various formulations (ethinyl estradiol,17‐beta estradiol, diethylstilbestrol, oestrone sulphate, quinesterol), as used in seven trials (Alvarez‐Sanchez 1996, Archer 2008, Boonkasemsanti 1996, Diaz 1990, Johannisson 1982, Said 1996 and Witjaksono 1996). Estrogen ‐ progestin combinations were used therapeutically in three trials (Alvarez‐Sanchez 1996, Sadeghi‐Bazargani 2006 and Witjaksono 1996). A progestin was used therapeutically in a single trial (Diaz 1990). Anti‐progestins were used therapeutically in two trials (Cheng 2000 and Weisberg 2006). An antiprogestin was combined with an estrogen as a method to treat irregular bleeding in two trials (Weisberg 2006, Weisberg 2009). A capillary protecting venotonic drug (registered in France as Cyclo 3, containing extract of Ruscus aculeatus with ascorbic acid and hesperidin methyl chalcone) was used therapeutically in one trial (Monteil‐Seurin 1985). In two trials (Subakir 2000 and d'Arcangues 2004), vitamin E was used therapeutically. Non‐steroidal anti‐inflammatory drugs (NSAIDs) were used therapeutically in eight trials (Archer 2008, Buasang 2009, Diaz 1990, Kaewrudee 1999, d'Arcangues 2004, Nathirojanakun 2006, Phaliwong 2004 and Tantiwattanakul 2004). The therapeutic effects of combined Vitamin E and NSAIDs were evaluated in a single trial (d'Arcangues 2004). A selective estrogen receptor modulator (SERM) was used therapeutically in one trial (Abdel‐Aleem 2005). An antifibrinolytic agent was used in two trials (Phupong 2006, Senthong 2009). A matrix metalloproteinase inhibitor (MMPI) alone was used in three trials (Abdel‐Aleem 2012, Weisberg 2006, Weisberg 2009). One trial evaluated the effects of combined MMPI and estrogen treatment ( Weisberg 2009); this same study also evaluated MMPI combined with an antiprogestin.

Four types of interventions were tested for their effects to prevent bleeding irregularities in users of progestin‐only contraceptives. These included estrogens alone (estradiol acetate, diethylstilbestrol, 17‐beta estradiol, quinesterol) as used prophylactically in five trials (Dempsey 2010, El‐Habashy 1970, Goldberg 2002; Madden 2012 and Parker 1980). Anti‐progestins were used prophylactically in four trials (Gemzell‐Danielsson 2002 , Jain 2003, and Massai 2004 and Warner 2010). The effects of vitamin C in preventing bleeding irregularities in new DMPA users were evaluated in a single four‐arm trial (Harel 2002). In addition, a trial of new LNG‐IUS users provided an NSAID for prophylaxis (Madden 2012).

All trials except (Dempsey 2010) were placebo‐controlled. Twenty four trials included a single intervention arm, while nine trials were designed to test the effects of more than one intervention, singly or in combination. In two trials (El‐Habashy 1970 and Parker 1980) the interventions tested, diethylstilbestrol and quinesterol, respectively, are no longer used in clinical practice, for any indication. In addition to the range of types of interventions, the treatments were used in different doses and according to different dosing schedules, and study participants were followed according to diverse schedules, for varying lengths of time.

Duration of Follow‐Up:
The duration of follow‐up varied among the trials, from less than three months to eighteen months. Most trials were designed to test the short‐term effects of the interventions, with only eight trials treating or following participants for longer than six months (Archer 2008, Parker 1980, Diaz 1990, Said 1996, Cheng 2000, Goldberg 2002, Massai 2004 and d'Arcangues 2004). The longest duration of follow‐up, as reported in two studies (Archer 2008, Said 1996) was one year and up to 18 months, respectively ; however, the long‐term results were not reported.

Outcome measures:
Although nearly all trials were designed to evaluate the effects of treatment on bleeding episodes as a primary outcome, the authors reported the outcome in different ways, including cessation of bleeding, number of bleeding days during treatment, number of bleeding days after treatment, number of days of treatment required before bleeding stopped, percent of women with bleeding free intervals of more than a given number of days, percent of women with acceptable bleeding, and others. Moreover, the data were presented in different ways including means, medians, and percentages. Other types of outcome measures included discontinuation of the treatment, discontinuation of the contraceptive method, side effects due to treatment, dissatisfaction with treatment and, in a few studies, blood loss during treatment.

Due to the lack of consistency in study design, outcome measures, and statistical reporting, very few studies could be combined in the analyses.

Risk of bias in included studies

This review included 33 trials. The methodological quality was assessed as mentioned above, in the methods section, with validity criteria scores based on allocation concealment, outcome assessment blinding and number/percent of study discontinuations. The overall assessment of risk of bias is shown in (Figure 1). According to these criteria, 10 of the trials were assigned "A" validity criteria scores with low risk of bias ( Abdel‐Aleem 2005,Archer 2008 , Buasang 2009 , Cheng 2000, Kaewrudee 1999, Nathirojanakun 2006, Phaliwong 2004 , Phupong 2006, Senthong 2009 and Weisberg 2006). In these trials, allocation concealment was considered to be adequate, outcome assessment blinding was achieved and fewer than 10% of the enrolled women discontinued prior to the end of the study(Figure 2). There were 12 trials that were considered to be at moderate risk of bias and were therefore given validity criteria scores of "B". In many of these studies, the allocation concealment was unclear ( Boonkasemsanti 1996, Jain 2003, Johannisson 1982, Massai 2004, Subakir 2000). The outcome assessment blinding was unclear or inadequate in three of these trials ( El‐Habashy 1970, Jain 2003, Johannisson 1982 ).In two trials the drop out rate was >10% but less than 15%(Abdel‐Aleem 2012,Weisberg 2009). Eleven trials were given a validity criteria score of "C", meaning that the study may be subject to a high risk of bias (Alvarez‐Sanchez 1996 , d'Arcangues 2004, Dempsey 2010 , Gemzell‐Danielsson 2002, Diaz 1990, Goldberg 2002,Madden 2012, Monteil‐Seurin 1985, Parker 1980 , Said 1996 and Witjaksono 1996).In all these trials at least two of the criteria were unmet or not clear or the drop out rate was markedly high. Overall, two thirds of the trials were of high to moderate quality , with 34% being considered "poor." The 10 trials (30% of all included studies) that were considered to be of good quality recruited 855 (23%) of 3,677 women enrolled in all included trials.

Figure 1

---

---

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 2

---

---

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Effects of interventions

1. Estrogen versus placebo (therapeutic treatment for LNG subcutaneous implant users; n=5 studies)
In five of the included studies, women reporting bleeding irregularities while using Norplant were administered an estrogen formulation and followed for resolution of symptoms (Alvarez‐Sanchez 1996, Archer 2008 , Boonkasemsanti 1996, Diaz 1990, and Witjaksono 1996). Together, these studies recruited 166 women in the respective estrogen treatment arms and 185 women in placebo groups; of these, data from 147 estrogen‐treated women and 173 placebo‐treated women were included in the analyses. Treatments included ethinyl estradiol (EE) tablets in four studies (Alvarez‐Sanchez 1996, Archer 2008 , Diaz 1990 and Witjaksono 1996) and an estradiol patch in the fifth (Boonkasemsanti 1996).

The efficacy of the various treatments was reported in different manners in these studies. In Alvarez‐Sanchez 1996, the primary outcomes included whether or not bleeding stopped within three days of the initiation of treatment and the number of bleeding days during the 20‐day treatment interval. Archer et al ( Archer 2008) reported the main study outcomes as being the mean duration (number of days) of bleeding and/or spotting during various intervals throughout the one‐month follow up period; however, the effect of EE treatment was inestimable, since standard deviations were not published. The Diaz 1990 study reported mean and total numbers of bleeding and spotting days during all treatment intervals throughout the entire year and did not separate individual treatment intervals. Women receiving estradiol reported shorter episodes of bleeding and spotting throughout the year and fewer days of bleeding and spotting. However, it is not clear how many treatment cycles were included in the analysis. But it is reported that the mean number of treatments per year per treated subject ranged from 2.2 to 3.1, with no differences between groups. Witjaksono 1996 reported on the number of bleeding or spotting days and episodes in 90‐day reference periods prior to and following the treatment, and the number of bleeding or spotting days per episode. The study results from the individual trials generally suggest that the estrogen treatment had a beneficial effect in stopping bleeding. Two of the included studies (Alvarez‐Sanchez 1996 and Boonkasemsanti 1996) reported the number of women who experienced irregular bleeding during treatment; according to our combined analysis, the use of estrogen reduced irregular bleeding during the treatment interval [RR 0.43, 95% CI 0.30‐0.61 ). In addition, the Alvarez‐Sanchez 1996 data support the conclusion that women receiving EE report a lower incidence of unacceptable bleeding patterns after treatment, as defined by the length of the post‐treatment bleeding‐free interval. Fewer women reported a bleeding‐free interval of less than 11 days after receiving the oral EE regimen than after receiving placebo [RR 0.28, 95% CI 0.14 ,0.58]. In contrast, no differences in the number of bleeding and/or spotting days within the 20 days following a 10‐day EE treatment period were reported by Archer et al (Archer 2008).

Based on our analysis of data reported by Alvarez‐Sanchez 1996, women receiving estrogen treatment appeared to be less likely to discontinue the use of Norplant for bleeding irregularities than were those receiving placebo, though the overall effect was not significant, as only one woman discontinued for this reason. Diaz 1990 did not report any method discontinuations for bleeding irregularities. Nausea or gastrointestinal discomfort was commonly reported in the former study: only two of 46 participants receiving placebo reported such side effects, whereas 17 of 43 receiving estrogen complained of gastric upset [RR 9.09, 95% CI 2.23, 37.06]. In the Diaz 1990 study, 5 of 13 study discontinuations in the estrogen group were associated with gastric intolerance, while none of the women in the placebo group discontinued from the study for this reason. Alvarez‐Sanchez 1996 reported discontinuation of treatment due to gastralgia or nausea in 2 of 50 women receiving treatment and none of the women in the placebo group. When the data for these studies were combined, it was clear that women receiving estrogen tended to be more likely to discontinue their participation in the trial due to side effects of the treatment than were women receiving placebo [RR 8.1, 95% CI 1.04, 63.40].

2. Estrogen versus placebo (therapeutic treatment for DMPA users; n= 1 study)
Only one of the included studies (Said 1996) was designed to evaluate the effects of an estrogen on bleeding irregularities in women using DMPA for family planning. This was one of the largest included studies, having randomised 278 DMPA users with irregular bleeding to receive either EE 50 µg, oestrone sulfate 2.5 mg or placebo for 14 days. This study was also designed to report on both short‐ and long‐term success, with follow‐up at 14 days (the duration of the treatment) and 12‐18 months; unfortunately, there was a high rate of discontinuation (over 40% in each group), giving the study a high risk of bias.

Ethinyl Estradiol was effective in stopping bleeding during treatment [RR 0.26, 95%CI 0.11, 0.60] and to a lesser extent after three months [RR 0.06, 95%CI 0.00, 1.00] . Oestrone sulphate was ineffective in stopping bleeding during treatment [RR 1.03, 95%CI 0.41, 2.59] and after three months.

The authors reported discontinuations and their causes among the study participants who presented at the study centres. But due to the high numbers of study participants who were lost to follow‐up and who dropped out of the study prematurely, without giving a reason, it was impossible to determine the true discontinuation rates, particularly as related to bleeding irregularities and side effects, respectively.

3. Estrogen versus placebo (prophylactic treatment for DMPA users; n= 4 studies)
Four studies were identified in which estrogen was administered to new acceptors of DMPA, with the goal of preventing or reducing bleeding irregularities. Two of these (El‐Habashy 1970 and Parker 1980) examined the effects of orally administered estrogens, diethylstilbestrol (DES) and quinesterol, respectively, the third (Goldberg 2002) evaluated the effects of a 17‐beta estradiol‐containing transdermal patch and the most recent study evaluated the use of a vaginal ring releasing estradiol acetate (Dempsey 2010).

Two of the studies reported efficacy of the treatment in terms that could be incorporated into the present analysis. El‐Habashy 1970 reported that women administered DES 1 mg daily exhibited more acceptable bleeding patterns (defined as bleeding episodes of less than 8 days) than women receiving placebo [RR 0.45, 95% CI 0.21, 0.96]. The trial of the vaginal ring ( Dempsey 2010 ) reported that the mean number of days of bleeding during treatment was reduced in women using the ring; however this finding was not statistically significant, with a weighted mean difference of ‐8.30 days in the treatment group (95% CI ‐20.2, 3.6). Parker 1980 reported the percent of women with acceptable bleeding patterns in terms of numbers of bleeding episodes throughout 3‐month intervals, following either one or three doses of quinestrol per DMPA injection segment (12 weeks). The primary outcome of the (Goldberg 2002) study was DMPA continuation rates and (Dempsey 2010 also reported on this outcome. The combined analysis demonstrated that estrogen treatment is ineffective as a means to improve DMPA continuation [RR 1.00, 95% CI 0.78, 1.28]. Three of the studies in this category suffered from very high discontinuation and non‐compliance rates, in both treatment and placebo groups

4. Estrogen versus placebo (therapeutic treatment for minipill users; n= 1 study)
One small (n=12) study (Johannisson 1982) was designed to evaluate the effects of estrogen treatment on bleeding irregularities experienced by minipill (300 µg norethisterone) users. The bleeding data could not be used in the current analysis, as confidence intervals, rather than standard deviations, were presented in the descriptions of bleeding during the treatment interval.

5. Combined estrogen and progestin versus placebo (therapeutic treatment for LNG subcutaneous implant users; n= 2 studies)
There were two studies (Alvarez‐Sanchez 1996; Witjaksono 1996) in which a combination of oral ethinyl estradiol and oral levonorgestrel was tested as a means to treat bleeding irregularities in Norplant users. The two studies together randomised 66 women to receive the treatment regimen and 64 to receive placebo; of these 61 and 60, respectively, were included in the analyses. As above, Alvarez‐Sanchez 1996 reported efficacy in terms of the percentage of women who experienced a cessation of bleeding within 3 days of treatment; the Witjaksono 1996 paper reported the number of bleeding days and episodes in 90‐day reference periods. According to the reports of the author and to our analyses, in the former study, the combined approach significantly reduced continued irregular bleeding during treatment, when compared with placebo [RR 0.08, 95% CI 0.03, 0.24] and reduced unacceptable bleeding after treatment (as defined by the number of women experiencing bleed‐free intervals of less than 11 days) [RR 0.02, 95% CI 0.00, 0.29].

Witjaksono 1996 did not report discontinuations or side effects in the study results; (Alvarez‐Sanchez 1996) reported that the discontinuation of Norplant due to bleeding disturbances was not significantly different between women in the combined hormone treatment and placebo groups, which was supported by our analysis [RR 0.33, 95% CI 0.01, 7.99]. Discontinuation due to the side effects of the treatment did not differ between groups; however, our analysis of the Alvarez‐Sanchez data revealed a higher incidence of reported side effects (nausea or gastralgia) in the treatment group than in the placebo group [RR 7.16, 95% CI 1.72, 29.71].

6. Combined estrogen and progestin versus placebo (therapeutic treatment for DMPA users; n= 1 study)
One small trial (Sadeghi‐Bazargani 2006 ) recruited DMPA users who had cessation of menstruation for two months. Twenty four women received low dose combined oral contraceptives (COCs) and 22 received placebo. Significantly fewer women in the treatment group than in the placebo group continued to experience amenorrhea [RR 0.38, 95% CI 0.19, 0.73]. The number of women who discontinued use of DMPA was less in the group assigned to receive low dose pills [RR 0.60, 95% CI 0.40, 0.88].

7. Progestin versus placebo (therapeutic treatment for LNG subcutaneous implant users; n= 1 study)
One study (Diaz 1990) evaluated the effect of administering an additional progestin to Norplant users experiencing bleeding episodes of eight or more days. These women received 0.03 mg levonorgestrel tablets, twice a day for 20 days, starting on the eighth consecutive day of bleeding. Women could treat extended bleeding up to five times in the year‐long study. The authors described a significant decrease in the total number of bleeding days and bleeding and spotting days over the entire year in women who ever used the levonorgestrel, compared with those who ever used the placebo, even when the untreated bleeding cycles were included in the analysis; however, the numbers of women in the final analysis were small (21 per group).

Three of the 47 women enrolled in the levonorgestrel group elected to have their Norplant system removed, though none for reasons related to bleeding; only one woman in the placebo group asked to discontinue the contraceptive method [RR 2.94, 95% CI 0.32, 27.21].

8. Antiprogestin versus placebo (therapeutic treatment for LNG subcutaneous implant users; n= 1 study)
One of the included studies (Cheng 2000) was designed to evaluate the effects of the anti‐progestin mifepristone on bleeding irregularities experienced by users of levonorgestrel‐containing subdermal contraceptive implants equivalent to Norplant. The study results stated that the women treated with mifepristone reported significantly shorter episodes of bleeding during the treatment than prior to the study; however, both the treatment and control groups demonstrated a decrease in the frequency of bleeding in the one year of follow up. In the current analysis, we found no significant difference between the groups with respect to the number of days of bleeding during treatment when reported 3 months after study initiation; at 6 months, however, women using the anti‐progestin reported fewer days of bleeding than did those receiving placebo [WMD ‐8.00, 95% CI ‐12.41, ‐3.59]. No difference in the number of days of bleeding after the treatment was noted. No women in either group discontinued the use of the contraceptive method due to bleeding irregularities. There was a significant effect of mifepristone to decrease patient dissatisfaction with the treatment [RR 0.46; 95% CI 0.23, 0.91].

9. Antiprogestin versus placebo (prophylactic treatment for LNG subcutaneous implant users; n= 1 study)
A single study (Massai 2004) was designed to evaluate the effect of two subsequent daily oral doses of 100 mg mifepristone in preventing bleeding irregularities in new users of Norplant. Of the 120 women enrolled, 116 completed 6 months of treatment and 115 were included in the analysis. The mean number of bleeding or spotting days during the course of the 180 days of the treatment period of the trial was significantly lower in women who had taken mifepristone for 2 days at 30 day intervals than in women given placebo [WMD ‐18.0; 95% CI ‐26.93, ‐9.07]. However, the improvement in bleeding patterns did not continue past the time of discontinuation of treatment and, by 6 months after treatment, there were no differences in reported acceptable bleeding patterns between the two groups. None of the women in the treatment group discontinued use of Norplant for continued irregular bleeding. Of importance, one woman who received the anti‐progestin did become pregnant during the course of the study. Two women in the placebo group elected to have their implants removed due to continued irregular bleeding. This difference was not found to be significant, nor was the difference between groups with respect to discontinuation from the study due to side effects of the treatments, in the current analysis.

10. Antiprogestin versus placebo (therapeutic treatment for Implanon users; n=1 study)
One pilot study (Weisberg 2006) randomised women who used Implanon implant and had bleeding irregularities to receive mifepristone (25 mg twice/day for one day) then placebo for four days or placebo only for five days. The authors reported that mifepristone alone was similar to placebo in stopping an episode of bleeding. Unfortunately, the results were presented in a way not suitable for our analysis. A similar proportion of women in both groups reported any side effect [0.93, 95% CI 0.61, 1.42] or, specifically, nausea and vomiting [0.64, 95% CI 0.23, 1.80].

11. Antiprogestin versus placebo (prophylactic treatment for DMPA users; n= 1 study)
Results of one small study (Jain 2003) suggest that a single tablet of 50 mg mifepristone, taken orally every 14 days, can lessen bleeding irregularities in new users of DMPA, in terms of percent days with breakthrough bleeding and percent cycles with bleeding intervals of more than 8 days.

12. Antiprogestin versus placebo (prophylactic treatment for minipill users; n= 1 study)
In one study (Gemzell‐Danielsson 2002), 103 women were randomised to take either 150 mg of a novel antiprogestogen (Org 31710) or placebo tablets once every 28 days, for a maximum of seven treatment cycles. Although the study authors report an improvement in the bleeding patterns of women who received the antiprogestin, our analysis did not reveal any differences between groups in terms of efficacy, discontinuation of the contraceptive method due to bleeding irregularities or side effects, or the incidence of side effects or adverse events.

13. Combined antiprogestin and estrogen versus placebo (therapeutic treatment for Implanon users; n= 1 study)
One pilot study (Weisberg 2006) and a larger follow‐on trial (Weisberg 2009 ) randomised women using Implanon and experiencing bleeding irregularities to take either mifepristone 50 mg for one day, followed by EE 20 µg/day for four days, or placebo for five days. The efficacy results were presented in a way not suitable for our analysis but the authors reported in both trials that mifepristone followed by EE was significantly more effective than placebo in stopping an episode of bleeding. Data extracted from the more recent trialWeisberg 2009 reflect the number of women who experienced continued irregular bleeding during treatment; none of the 40 women randomized to receive the active treatment reported continued bleeding, compared with 11 of the 37 women who received placebo [RR 0.04, 95% CI 0.00, 0.66]. The analysis of the pilot study demonstrated no significant differences between the two groups of women in reporting any side effects [RR 1.22, 95% CI 0.84, 1.75]. The 2009 trial reported no difference between treatment groups with respect to the number of women who experienced headache; a combined analysis of both trials revealed no effect of treatment group on the number of women experiencing nausea or vomiting [RR 0.5, 95% CI 0.18, 1.38]. The more recent trial reported on discontinuation of the contraceptive method at 90 and 180 days following initiation of treatment. Treatment with mifepristone followed by EE did not result in a favourable effect on method continuation rates at either time point.

14. Venotonic versus placebo (therapeutic treatment for minipill users; n= 1 study)
In one of the included studies (Monteil‐Seurin 1985), women using a norethisterone acetate‐containing minipill and complaining of bleeding irregularities were administered either a capillary protecting venotonic drug (registered in France as Cyclo 3, containing extract of Ruscus aculeatus with ascorbic acid and hesperidin methyl chalcone) or placebo. The treatment schedule was cyclic, with drug exposure for 20 continuous days, followed by a drug‐free period during the expected time of menses. The mean number of bleeding days during treatment was lower in the group using the venotonic drug in comparison to the placebo group [WMD ‐2.0, CI ‐2.78, ‐1.22]. This was the only one of the included studies that attempted to quantify the amount of blood lost during treatment, as a listed endpoint, and the study authors indicated that women who received the Cyclo 3 needed less menstrual protection than did women in the placebo group. There were no study drop outs reported and no data were presented on side effects due to the treatment. According to the study authors, treated women subjectively reported less discomfort due to bleeding after 3 treatment cycles than did women who received placebo.

15. Vitamin E versus placebo (therapeutic treatment for LNG subcutaneous implant users; n= 2 studies)
Two of the included studies (d'Arcangues 2004 and Subakir 2000) were conducted to evaluate the effectiveness of vitamin E, 200 mg/day for 10 days, in the treatment of bleeding irregularities experienced by women using Norplant implants. In Subakir study, the mean number of bleeding days after the first treatment cycle was significantly less among the 38 women taking vitamin E than among the 34 women assigned to the placebo group [WMD ‐4.4, CI ‐5.02, ‐3.78]. However, in the second, larger study (d'Arcangues 2004), with more than 100 women in each group, there were no statistically significant differences between the groups in terms of the length of bleeding/spotting episodes and the length of bleeding‐free intervals that followed cessation of bleeding. d'Arcangues 2004 reported median numbers of days of bleeding, rather than means, given the non‐normal distribution of the data. In this study, discontinuation rates were high but did not differ between groups, nor did the incidence of side effects.

16. Non‐steroidal anti‐inflammatory drug versus placebo (therapeutic treatment for LNG subcutaneous implant users; n=5 studies)
Five of the included studies ( Archer 2008 , Buasang 2009 , d'Arcangues 2004 , Diaz 1990 and Kaewrudee 1999) were designed to evaluate the effects of non‐steroidal anti‐inflammatory drugs (NSAIDS) on bleeding irregularities experienced by women using LNG‐releasing sub‐cutaneous implants (either Norplant or Jadelle). The regimens included oral ibuprofen 800 mg, three times/day for five days (Diaz 1990); oral ibuprofen 800 mg, two times/day for five days ( Archer 2008); oral mefenamic acid 500 mg, twice daily for 5 days (Kaewrudee 1999); and oral aspirin, 80 mg/day for 10 days (d'Arcangues 2004) and celecoxib 200 mg/day for five days (Buasang 2009).

The earliest paper reported that women taking ibuprofen experienced a decrease in the mean number of bleeding days after initiating treatment in all treated bleeding intervals over the course of a year and fewer bleeding and spotting days during the year (Diaz 1990). Of data from Kaewrudee and Buasang , 14 of the 54 women taking an NSAID continued to experience irregular bleeding during the treatment, compared with 44 of 53 women in the placebo group [RR 0.32; 95% CI 0.2, 0.51]. In addition, Kaewrudee et al reported that women randomized to the mefenamic acid group experienced fewer days of bleeding during the treatment [WMD ‐5.6; 95% CI ‐10.04, ‐1.16]. The percentage of women with unacceptable bleeding after treatment was significantly less in the treated group as reported by both( Kaewrudee 1999 and Buasang 2009 )and confirmed in the present combined analysis [RR 0.38; 95% CI 0.25, 0.6]. In the d'Arcangues 2004 trial, the median duration of bleeding/spotting episodes after treatment and the median lengths of the bleed‐free intervals were not significantly different in the drug and placebo groups, as reported by the authors. Archer et al (Archer 2008) reported the main study outcomes as being the mean duration (number of days) of bleeding and/or spotting during various intervals throughout the one‐month follow up period; however, the effect of ibuprofen treatment was inestimable, since standard deviations were not published. No differences in the number of bleeding and/or spotting days within the 25 days following a five‐day ibuprofen treatment period were reported by these authors.

In the Diaz 1990 trial none of the women receiving ibuprofen requested removal of the Norplant system, while one woman in the placebo group requested removal, though not for bleeding‐associated reasons. No differences in discontinuation rates or incidence of side effects (headache and gastrointestinal upset) were noted in the single studies or in the combined analyses (where combinations were feasible). Patient dissatisfaction with treatment was significantly lower as reported by Jadelle users taking celecoxib for bleeding irregularities, in comparison with the placebo group ( Buasang 2009) [RR 0.29; 95% CI 0.11, 0.72].

17. Combined non‐steroidal anti‐inflammatory drug and vitamin E versus placebo (therapeutic treatment for LNG subcutaneous implant users; n= 1 study)
In the large study reported in d'Arcangues 2004, over 120 women using Norplant were randomly assigned to receive a combination of aspirin and vitamin E, while the same number received placebo. The efficacy data were reported as medians, and no benefit of the treatment was demonstrated in the original analysis. In addition, our analyses indicated that there were no differences between the groups with respect to study discontinuation or side effects related to the treatment .

18. Non‐steroidal anti‐inflammatory drugs versus placebo (therapeutic treatment for DMPA users; n= 2 studies)
Two trials (Tantiwattanakul 2004 and Nathirojanakun 2006) evaluating the therapeutic effect of an NSAID on bleeding patterns in women using DMPA were included in the analysis. In the first, women were randomised to receive mefenamic acid (n=23) or placebo (n=25). In the second trial, 46 women were randomised to receive either valdecoxib or placebo. . In a combined analysis, fewer women receiving NSAID treatment continued bleeding during or just after the treatment period, versus in the placebo group [RR 0.42; 95% CI 0.25, 0.72]. In the first trial gastrointestinal discomfort was reported by six women in the treatment group and by none in the control group; this difference was not found to be significant in our analysis. In the second trial no adverse effects were reported.

19. Selective estrogen receptor modulator versus placebo (therapeutic treatment for LNG subcutaneous implant users; n= 1 study)
Abdel‐Aleem 2005 was the only included trial to examine the efficacy of a selective estrogen receptor modulator (SERM), in this case tamoxifen, on bleeding irregularities experienced by women using progestin‐only methods of fertility regulation. The study was limited to women using Norplant. women received tamoxifen, 10 mg tablets twice daily for 10 days or placebo. Women randomised to the treatment arm were significantly less likely to experience irregular bleeding during treatment [RR 0.41; 95% CI 0.21, 0.80] or unacceptable bleeding after treatment [RR 0.38; 95% CI 0.18, 0.88] or to discontinue the use of Norplant for any reason [RR 0.20; 95% CI 0.05, 0.87]. The effect of the SERM in reducing Norplant discontinuation for bleeding‐related reasons bordered on significance [RR 0.22; 95% CI 0.05, 0.98]. Women receiving tamoxifen were more likely to report satisfaction with their treatment regimen [RR 0.22; 95% CI 0.11, 0.45] and were no more likely to report side effects.

20. Antifibrinolytic versus placebo (therapeutic treatment for LNG subcutaneous implant users; n= 1 study)
One randomised trial (Phupong 2006) evaluated the effects of an antifibrinolytic agent on bleeding patterns in women using Norplant and reporting irregular bleeding. Sixty‐eight Norplant users were enrolled; half received tranexamic acid 500 mg twice/day for five days and the other half received similar placebo. The percentage of women whose irregular bleeding stopped at 7 days after the initiation of treatment was significantly higher in the tranexamic acid group than the placebo group [RR 0.55, 95 % CI 0.32, 0.92]. The percentage of women with unacceptable bleeding by four weeks after the initiation of treatment was higher in the treatment group (41%) than in the control group (24%) [RR 1.75, 95% CI 0.85, 3.62].

21. Matrix metalloproteinase inhibitor versus placebo (therapeutic treatment for Implanon users; n= 2 studies)
In a pilot study (Weisberg 2006) and a larger follow‐on trial (Weisberg 2009), Implanon users who experienced prolonged or frequent bleeding patterns were randomised to receive doxycycline 100 mg/twice daily or similar placebo, for five days. Follow up was for a period of 90 days in the pilot study; the larger trial implemented a 90‐day treatment period, then a 90 day follow‐up period. In the earlier study, the authors reported that doxycycline was more effective than placebo in stopping an episode of bleeding; this finding was not replicated in the larger trial. Data extracted from the 2009 trial showed no treatment benefit with respect to continued irregular bleeding during treatment; 11 of the 32 women randomized to receive the active treatment reported continued bleeding, compared with 11 of the 37 women who received placebo [RR 1.16, 95% CI 0.58, 2.3].

At 90 and 180 days following initiation of treatment, there were no differences in Implanon discontinuation rates between women receiving doxycycline and those receiving placebo [RR 0.98; 95% CI 0.26, 3.64] and [RR 0.76; 95% CI 0.31, 1.84], respectively.

There were no differences in the reporting of any side effects between the two groups [RR 0.83, 95% CI 0.53, 1.29]. The side effect of nausea/vomiting was reported in both trials; a combined analysis demonstrated no benefit of the treatment, though results were inconsistent between the studies. In the 2009 trial, fewer women reported headache in the treatment group; however, this finding was not significant.The authors reported that there was good compliance with treatment intake, with only 10% of study participants reporting having missed any tablet.

.

22. Matrix metalloproteinase inhibitor versus placebo (therapeutic treatment for DMPA users; n=1 study)

In a recently published trial (Abdel‐Aleem 2012), 68 women using DMPA who experienced prolonged or frequent bleeding and/or spotting patterns were randomised to receive doxycycline 100 mg/twice daily or an identical placebo, for five days. Trial participants were followed for three months after the initial treatment to compare patterns of bleeding and spotting. When compared with a placebo, doxycycline treatment showed no benefit in stopping an episode of bleeding within 10 days of treatment (RR 0.88, 95% CI 0.64, 1.21). Moreover, during the follow‐up of two cycles post treatment, no differences in either the mean number of bleeding or spotting days recorded by study participants were observed. Two women in doxycycline arm (2/34) and three women taking placebo (3/34) discontinued using DMPA during the trial (RR 0.67, 95% CI 0.12, 3.74). Several women reported side effects due to their treatments: in the doxycycline arm, two women experienced nausea and one complained of diarrhoea; whereas one woman who received placebo reported nausea.

23. Antifibrinolytic versus placebo (therapeutic treatment for DMPA users; n= 1 study)

One randomized trial (Senthong 2009) evaluated the effects of tranexamic acid for treatment of irregular bleeding secondary to use of DMPA. One hundred women were enrolled and randomized to receive either tranexamic acid (1000 mg per day for five days) or placebo, and were followed for four weeks after the initial treatment. Current analysis of the study results showed that six out of the 50 women who were assigned the treatment group, and 45 of the 49 women in the placebo group, experienced continued irregular bleeding during the first week of treatment [RR 0.13, 95% CI 0.06, 0.28]. There was also a significant effect of tranexamic acid on unacceptable bleeding after treatment; women in the treatment group experienced fewer days of bleeding and spotting in the four week follow‐up period [mean number of days: 5.7 ± 2.5 days vs 17.5 ± 7.2 days, respectively].

24. Non‐steroidal anti‐inflammatory drug versus placebo (therapeutic treatment for Implanon users; n= 1 study)

A single trial (Phaliwong 2004) evaluated the effects of mefenamic acid for treatment of irregular bleeding secondary to use of Implanon. Fifty women were enrolled and randomized to receive either 1500 mg mefenamic acid per day for five days or placebo. They were followed for four weeks after the initial treatment. Eight out of 23 women who were assigned to the treatment group, and 18 of the 23 women in the placebo group, experienced continued irregular bleeding during the first week following initiation of treatment [RR 0.44, 95% CI 0.24, 0.81, current analysis]. Duration of bleeding and spotting during the 28 days of follow‐up was reported by the authors to be significantly less among women taking mefenamic acid compared with women assigned to the placebo group [mean number of days: 10.5 days vs 16.8 days, respectively]; however, we were unable to estimate the mean difference in continued irregular bleeding because the standard deviations were not published.

Due to the small number of women enrolled in the trial, and wide confidence intervals surrounding the point estimate of effect, in the current analysis of the results, there were no observed differences in side effects experienced by women taking mefenamic acid compared with placebo. Abdominal discomfort was reported by three women in the treatment group, compared with none in the placebo group. Headache was reported by one woman in each group and breast tenderness was experienced by two women in the mefenamic acid group, but none in the placebo group.

25. Antiprogestin versus placebo (prophylactic treatment for LNG IUS users; n= 1 study)

One randomized trial (Warner 2010) evaluated the prophylactic effect of the use of an antiprogestin to reduce bleeding irregularities among new users of the LNG IUS. 136 women were enrolled; 69 were randomized to receive CDB‐2914 (150 mg per day for three days, starting on days 21, 49 and 77 after device insertion) and 67 received placebo. Follow up visits were scheduled at 1, 3 and 6 months after device insertion. According to current analysis,during the first 28 days following the initial treatment (i.e. initiation on day 21 following LNG IUS insertion), there was a significant difference in the percentage of days in which women experienced bleeding and/or spotting. Women taking anti‐progestins reported, on average, bleeding and/or spotting on 46.7% ± 20.6 days during the 28 day period, whereas women in the placebo group recorded irregular bleeding on 57.3% ± 24.8 days during the first 28‐day cycle. However, any beneficial effect of treatment with this antiprogestin disappeared by the third treatment cycle (i.e. the 28 days following initiation of treatment on day 77 after insertion). Women taking antiprogestin reported, on average, bleeding and/or spotting on 38.2% ± 23.7 days during the 28 day period, whereas women in the placebo group recorded irregular bleeding on 28.7% ± 20.7 days during the third 28 day cycle. In fact, women in the placebo group reported a significantly lower percentage of days of bleeding and/or spotting during the third cycle.

Twenty‐three percent (14/61) of women taking CDB‐2914 reported a tendency to gain weight during the trial, compared with only 4% (2/25) of women in the control group. While this difference is statistically significant, according to patient characteristics collected at baseline, women who were randomized to the treatment group were more likely to have discontinued use of a previous contraceptive method due to weight gain.

There was no significant difference with regards to patient dissatisfaction with treatment, as implied by drug noncompliance. Further reports of blood loss improvement did not differ between the two groups. Similar numbers of women in each group requested removal of their LNG IUS during the trial.

26. Vitamin C supplementation versus placebo (prophylactic treatment for DMPA users; n= 1 study)

A single trial (Harel 2002) evaluated the effects of oral vitamin C supplementation on bleeding patterns among adolescents initiating DMPA use. Fifty‐five adolescent girls (age 16 ± 1 years) were randomly assigned to one of four groups (three treatment groups and a placebo group); only one of the treatment groups was included as a possible prophylactic treatment for bleeding irregularities, the other two treatment groups were not included in the present analysis. The treatment group received 500 mg vitamin C plus placebo per day and the placebo group received 2 placebo tablets per day. Follow‐up visits were scheduled at three and six months after the initial injection. There was no significant difference in the number of days of bleeding during the follow up period experienced by women in the treatment group, versus the placebo.

27. Antiprogestin and matrix metalloproteinase inhibitor versus placebo (therapeutic treatment for Implanon users; n= 1 study)

In a single trial (Weisberg 2009), Implanon users who experienced prolonged or frequent bleeding patterns were randomised to receive doxycycline 100 mg/twice daily for five days, with the addition of 25 mg mifepristone on the first day, or similar placebo. The trial implemented a 90‐day treatment period, with a 90 day post‐treatment follow‐up period. Extracted data demonstrated a significant treatment benefit with respect to continued irregular bleeding during treatment; 1 of the 35 women randomized to receive the active treatment reported continued bleeding, compared with 11 of the 37 women who received placebo [RR 0.10; 95% CI 0.01, 0.71].

At 90 and 180 days following initiation of treatment, there were no differences in Implanon discontinuation rates between women receiving doxycycline and mifepristone and those receiving placebo, despite a tendency for improved continuation rates at 180 days in the treatment group [RR 0.49; 95% CI 0.09, 2.52] and [RR 0.33; 95% CI 0.09, 1.11], respectively.

The side effect of nausea/vomiting was reported slightly more frequently by women in the treatment group; however, this difference was not significant. Alternatively, fewer women in the treatment group reported headaches; again, this finding was not significant.

28. Estrogen and matrix metalloproteinase inhibitor versus placebo (therapeutic treatment for Implanon users; n= 1 study)

In a single trial (Weisberg 2009), Implanon users who experienced prolonged or frequent bleeding patterns were randomised to receive doxycycline 100 mg/twice daily combined with EE 20 μg/day for five days, or similar placebo. The trial implemented a 90‐day treatment period, with a subsequent 90 day follow‐up period. Extracted data demonstrated no treatment benefit with respect to the number of women reporting continued irregular bleeding during treatment; 14 of the 35 women randomized to receive the active treatment reported continued bleeding, compared with 11 of the 37 women who received placebo [RR 1.35; 95% CI 0.71, 2.55].

At 90 and 180 days following initiation of treatment, there were no differences in Implanon discontinuation rates between women receiving doxycycline and EE and those receiving placebo.

The number of women reporting the side effects of nausea/vomiting or headache was not different between treatment groups.

29. Estrogen versus placebo (prophylactic treatment for LNG IUS users; n= 1 study)

One recently published trial (Madden 2012) evaluated whether the application of a transdermal 0.1 mg estradiol patch decreased bleeding and spotting among new users of the LNG IUS. Overall, the trial enrolled 129 women initiating the use of the LNG IUS and randomly assigned them to three arms for twelve weeks of treatment (84 days). Forty‐four women were assigned to the transdermal estradiol arm, 42 women received a 500 gm NSAID, and 43 women received a placebo identical to the NSAID. Comparisons between the NSAID and placebo arms are presented and analysed elsewhere in this review.

Women assigned to the transdermal estradiol group applied their patch the day following IUS insertion and used it continuously, changing the patch on a weekly basis, for the 12 week treatment period. Beginning the day after IUS insertion, the placebo group took their study medication twice daily for the first five days of each four week period during the 12 weeks of treatment. Trial participants were asked to complete study diaries to record the occurrence of bleeding and spotting as well as compliance with the study's protocol. Telephone follow‐up was conducted at 4, 8 and 16 weeks, and an in‐person follow‐up visit was scheduled at 12 weeks.

Because the results were reported as medians , range and quartiles, current analysis could not be done , and the results presented as reported by the authors.During the entire treatment period of the trial, women using the estradiol patch reported a greater number bleeding/spotting days than the placebo group [median and range: 44 days (2‐82 days) for estradiol patch arm versus 32 days (9‐84 days) for placebo arm], however the difference was not statistically different (p‐value = 0.15). The distribution of bleeding and spotting over the 12‐week treatment period showed that women in the estradiol patch arm were significantly more likely to experience patterns of bleeding/spotting within the highest quartile (i.e., 54‐84 days) compared with women who received placebo (40.9% versus 18.l %, p‐value = 0.02). During the four week post‐treatment period, there were no reported differences in bleeding/spotting between women in the estradiol patch or placebo groups.

Over the 12 treatment period, one woman in each group discontinued the use of the LNG‐IUS due to dissatisfaction with the demands of the trial (patch arm) and IUS expulsion (placebo arm). One woman in the patch group discontinued her treatment due to headache complaints. A greater proportion of women in the estradiol patch arm reported dissatisfaction with their treatment compared with the placebo group both at 4 weeks (39.5% versus 11.6%, p‐value=0.01) and 12 weeks follow‐up (13.5% versus 0).

30. Non‐steroidal anti‐inflammatory drug versus placebo (prophylactic treatment for LNG IUS users; n= 1 study)

A recently published trial (Madden 2012) evaluated the prophylactic effect of naproxen in decreasing bleeding and spotting irregularities among new users of the LNG IUS. The trial enrolled 129 new users of the LNG IUS, and randomly assigned them to three arms for twelve weeks of treatment (84 days). Forty‐two women were assigned to receive naproxen (500 gm), 43 women received a placebo identical to naproxen, and 44 women were assigned to a transdermal estradiol arm. Comparisons between the estradiol patch and the placebo arms are presented and analysed elsewhere in this review.

Beginning the day after IUS insertion, the naproxen and placebo group took their 500 gm study medication twice daily for the first five days of each four week period during the 12 weeks of treatment. Trial participants were asked to complete study diaries to record the occurrence of bleeding and spotting as well as compliance with the study's protocol. Telephone follow‐up was conducted at 4, 8 and 16 weeks, and an in‐person follow‐up visit was scheduled at 12 weeks.

Because the results were presented as means, range and quartiles, current analysis could not be done per review methodology. The authors reported that over the entire treatment period, women who received naproxen reported fewer bleeding/spotting days than the placebo group [median and range: 27.5 days (5‐83 days) naproxen arm versus 32 days (9‐84 days) placebo arm], however the difference was not statistically different (p‐value = 0.1). The distribution of bleeding and spotting over the 12‐week treatment period showed that women taking naproxen were more likely to experience patterns of bleeding/spotting within the lowest quartile (i.e., 2‐21 days) compared with women who received placebo (42.9% versus 16.3 %). During the four week post‐treatment period, there were no reported differences in bleeding/spotting between women in either group.

Over the 12 week treatment period, two women who received naproxen discontinued use of the LNG IUS due to pregnancy and chest pain. A woman in the placebo group discontinued using her LNG‐IUS due to IUS expulsion. Reports of dissatisfaction with treatment differed minimally between the naproxen or placebo groups at 4 weeks (9.5% versus 11.6%); however, more women taking naproxen were dissatisfied with the treatment at 12 weeks (11.4% versus 0).