Progestin-only pills for contraception (Review)

Transcription

1 Grimes DA, Lopez LM, O Brien PA, Raymond EG This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library 2013, Issue 11

2 T A B L E O F C O N T E N T S HEADER ABSTRACT PLAIN LANGUAGE SUMMARY BACKGROUND OBJECTIVES METHODS RESULTS DISCUSSION AUTHORS CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES CHARACTERISTICS OF STUDIES DATA AND ANALYSES Analysis 1.1. Comparison 1 Desogestrel 75 µg versus levonorgestrel 30 µg daily, Outcome 1 Pregnancy Analysis 1.2. Comparison 1 Desogestrel 75 µg versus levonorgestrel 30 µg daily, Outcome 2 Discontinuation because of adverse events Analysis 1.3. Comparison 1 Desogestrel 75 µg versus levonorgestrel 30 µg daily, Outcome 3 Discontinuation because of irregular bleeding Analysis 1.4. Comparison 1 Desogestrel 75 µg versus levonorgestrel 30 µg daily, Outcome 4 Discontinuation for all reasons Analysis 2.1. Comparison 2 Mifepristone 5 mg versus levonorgestrel 30 µg daily, Outcome 1 Pregnancy Analysis 2.2. Comparison 2 Mifepristone 5 mg versus levonorgestrel 30 µg daily, Outcome 2 Amenorrhea Analysis 2.3. Comparison 2 Mifepristone 5 mg versus levonorgestrel 30 µg daily, Outcome 3 Mean endometrial thickness at 24 weeks Analysis 2.4. Comparison 2 Mifepristone 5 mg versus levonorgestrel 30 µg daily, Outcome 4 Endometrial biopsy to evaluate endometrial cavity >12 mm on ultrasound Analysis 3.1. Comparison 3 Ethynodiol diacetate 0.25 mg versus ethynodiol diacetate 1.0 mg plus mestranol 0.1 mg, Outcome 1 Irregular cycles Analysis 3.2. Comparison 3 Ethynodiol diacetate 0.25 mg versus ethynodiol diacetate 1.0 mg plus mestranol 0.1 mg, Outcome 2 Intercycle bleeding Analysis 3.3. Comparison 3 Ethynodiol diacetate 0.25 mg versus ethynodiol diacetate 1.0 mg plus mestranol 0.1 mg, Outcome 3 Discontinuation from trial Analysis 4.1. Comparison 4 Norethisterone acetate 0.3 mg versus norgestrel mg, Outcome 1 Pregnancy Analysis 4.2. Comparison 4 Norethisterone acetate 0.3 mg versus norgestrel mg, Outcome 2 Continuation at one year Analysis 4.3. Comparison 4 Norethisterone acetate 0.3 mg versus norgestrel mg, Outcome 3 Discontinuation because of menstrual disturbance Analysis 4.4. Comparison 4 Norethisterone acetate 0.3 mg versus norgestrel mg, Outcome 4 Discontinuation because of other side effects Analysis 4.5. Comparison 4 Norethisterone acetate 0.3 mg versus norgestrel mg, Outcome 5 Discontinuation for reasons unconnected with treatment ADDITIONAL TABLES APPENDICES WHAT S NEW HISTORY CONTRIBUTIONS OF AUTHORS DECLARATIONS OF INTEREST SOURCES OF SUPPORT INDEX TERMS i

3 [Intervention Review] David A Grimes 1, Laureen M Lopez 2, Paul A O Brien 3, Elizabeth G. Raymond 4 1 Obstetrics and Gynecology, University of North Carolina, School of Medicine, Chapel Hill, North Carolina, USA. 2 Clinical Sciences, FHI 360, Research Triangle Park, North Carolina, USA. 3 Westside Contraceptive Services, London Community Healthcare, London, UK. 4 Gynuity Health projects, New York, New York, USA Contact address: David A Grimes, Obstetrics and Gynecology, University of North Carolina, School of Medicine, CB#7570, Chapel Hill, North Carolina, , USA. david_grimes@med.unc.edu. Editorial group: Cochrane Fertility Regulation Group. Publication status and date: Edited (no change to conclusions), published in Issue 11, Review content assessed as up-to-date: 29 October Citation: Grimes DA, Lopez LM, O Brien PA, Raymond EG.. Cochrane Database of Systematic Reviews 2013, Issue 11. Art. No.: CD DOI: / CD pub3. Background A B S T R A C T The introduction of a new progestin-only oral contraceptive in Europe has renewed interest in this class of oral contraceptives. Unlike the more widely used combined oral contraceptives containing an estrogen plus progestin, these pills contain only a progestin (progestogen) and are taken without interruption. How these pills compare to others in their class or to combined oral contraceptives is not clear. Objectives This review examined randomized controlled trials of progestin-only pills for differences in efficacy, acceptability, and continuation rates. Search methods Through October 2013, we searched the computerized databases MEDLINE, Cochrane Central Register of Controlled Trials (CEN- TRAL), POPLINE, and LILACS for studies of progestin-only pills. We also searched for current trials via ClinicalTrials.gov and ICTRP. Previous searches also included EMBASE. Selection criteria We included all randomized controlled trials in any language that included progestin-only pills for contraception. We incorporated any comparison with a progestin-only pill; this could include different doses, other progestin-only pills, combined oral contraceptives, or other contraceptives. Data collection and analysis The first author abstracted the data and entered the information into RevMan 5. Another author performed a second, independent data abstraction to verify the initial data entry. We attempted to extract life-table rates (actuarial or continuous) and used the rate difference as the effect measure. Where life-table rates were not published, we used the incidence rate ratio (ratio of Pearl rates). Where only the crude number of events was published, we calculated the Peto odds ratio with 95% confidence interval (CI) using a fixed-effect model. For continuous variables, the mean difference (MD) was computed with 95% CI. Because of disparate exposures, we were not able to combine studies in meta-analysis. 1

4 Main results Six trials met the inclusion criteria. We have not found any new studies since the initial review. In the trial comparing the desogestrel versus levonorgestrel progestin-only pill, desogestrel was not associated with a significantly lower risk of accidental pregnancy; the rate ratio was 0.27 (95% CI 0.06 to 1.19). However, the desogestrel progestin-only pill caused more bleeding problems, although this difference was not statistically significant. The trial comparing low-dose mifepristone versus a levonorgestrel progestin-only pill found similar pregnancy rates. In the trial comparing ethynodiol diacetate versus a combined oral contraceptive, irregular cycles occurred in all women assigned to the progestin-only pill (odds ratio ; 95% CI 7.61 to ). In a trial comparing two progestin-only and two combined oral contraceptives, the progestin-only pill containing levonorgestrel 30 µg had higher efficacy than did the pill containing norethisterone 350 µg. An early trial found megestrol acetate inferior to other progestin-only pills in terms of efficacy. A study of the timing of pill initiation after birth found no important differences, but high losses to follow up undermined the trial. Authors conclusions Evidence is insufficient to compare progestin-only pills to each other or to combined oral contraceptives. P L A I N L A N G U A G E S U M M A R Y Progestin-only pills, as their name implies, contain just one hormone. The more common birth control pills combine two hormones. How these one-hormone pills compare to each other or to two-hormone pills is not clear. Hence, we did this review to compare progestin-only pills to other similar pills or to combined (two-hormone) pills. Through October 2013, we did a computer search and literature search to find randomized trials of progestin-only pills. We found six trials for the initial review. We have not found any more studies since then. Some studies are several decades old and not very relevant to pills available today. A newer pill containing the progestin desogestrel may work better to prevent pregnancy than an older pill with levonorgestrel. The newer pill caused more bleeding problems. Pills with levonorgestrel may be more effective than pills with other progestins that are no longer used. These studies are not adequate to tell how progestin-only pills compare to each other or to combined (two-hormone) pills. Larger studies with currently used pills are needed to answer these questions. B A C K G R O U N D The introduction of a new progestin-only oral contraceptive in Europe (FFPRHC 2003; Korver 2005) renewed interest in this class of oral contraceptives. Unlike the more widely used combined oral contraceptives containing an estrogen plus progestin (Edelman 2005; van Vliet 2011; Lopez 2012), these pills contain only a progestin (progestogen) and are taken without interruption. Progestin-only pills (POPs) have both advantages and disadvantages when compared with combined pills. The pill-taking regimen is simple and fixed; no pill color changes or days without pill-taking occur. Fertility returns promptly upon discontinuation. These pills are appropriate for women who cannot or should not take estrogen in combined oral contraceptives, for example, a woman older than 35 years who smokes cigarettes (ACOG 2006). They may also appeal to women who want to use the lowest effective dose of steroids to provide contraception. Some progestinonly pills were designed to allow ovulation and regular menses, although that goal has been reversed with the desogestrel pill. Lacking estrogen, progestin-only pills may have a lower risk of complications. A World Health Organization case-control study found no significant increase in the risk of stroke, myocardial infarction, and venous thromboembolism among users of progestin-only pills compared with non-users (WHO 1998). A cohort study from Denmark also found no statistically significant association between progestin-only pills and venous thromboembolism (Lidegaard 2009). Although the literature is unsettled concerning the potential impact of combined oral contraceptives on lactation, no concerns exist for progestin-only pills (Moggia 1991; Dunson 1993; McCann 1994; Bjarnadóttir 2001; FFPRHC 2004). No 2

5 data are available concerning a potential effect on infant brain or liver. Like depot-medroxyprogesterone acetate injections for contraception (Manchikanti 2007), progestin-only pills may reduce the frequency of sickle cell crises. Less restrictive screening for use may facilitate wider distribution by non-clinicians or over-thecounter provision. Disadvantages include the recommendation for careful compliance and the disruption of normal menstrual patterns (FSRH 2008). These changes include irregular bleeding, short or long cycles, bleeding and spotting, prolonged bleeding, or no bleeding at all. Overall, progestin-only pills are associated with more days of bleeding and spotting than are combined oral contraceptives (Raymond 2011). The timing of ingestion of progestin-only pills may be more important than with combined oral contraceptives. However, the potential effect of timing on efficacy has not been adequately studied, and, for practical reasons, likely will not be studied. Inter-individual variation in progestin metabolism rates is wide (Goldzieher 1994; Wallach 2000). Progestin-only pills may not provide as much protection against ectopic pregnancy as do combined oral contraceptives (McCann 1994), and they may be associated with more functional ovarian cysts than are combined oral contraceptives (Raymond 2011). Why it is important to do this review How different progestin-only pills compare to each other in terms of efficacy and continuation is unclear. Because of the lower dose of progestin than in combined oral contraceptives, the reliance on mechanisms other than ovulation inhibition, and the potential importance of consistent daily pill taking, progestin-only pills may not be as effective as combined oral contraceptives. Stated alternatively, progestin-only pills may be less forgiving of late or missed pills. However, data to support or refute this hypothesis are sparse. O B J E C T I V E S This review examined randomized controlled trials of progestinonly pills for differences in efficacy, acceptability, and continuation rates. M E T H O D S Criteria for considering studies for this review Description of the intervention Commercially available progestin-only pills contain low doses of levonorgestrel, norethindrone (norethisterone), ethynodiol diacetate, or desogestrel. The progestin dose is lower than that in combined oral contraceptives, hence the common name mini-pill (Wallach 2000; Raymond 2011). Types of studies We included all randomized controlled trials in any language that included progestin-only pills for contraception. Types of participants Women requiring contraception with data in the eligible trials were included in the review. How the intervention might work Progestin-only pills prevent pregnancy through several potential mechanisms. Ovulation is variably inhibited (FSRH 2008), and the cervical mucus becomes thick and relatively impenetrable to sperm. Ciliary action in the fallopian tubes is altered, as is the histology of the endometrium, potentially making implantation less likely (McCann 1994; Wallach 2000; Raymond 2011). The effectiveness of progestin-only pills is not well-established. Reported pregnancy rates in the first year of use range from zero to 13% (Raymond 2011; Trussell 2011); some authors have estimated that typical pregnancy rates may approximate those of combined oral contraceptives (Wallach 2000). Consistent timing is considered important for successful use, but supporting evidence is limited. In general, pregnancy rates may also be influenced by the woman s age, frequency of coitus, body mass, and lactation. Types of interventions We included any comparison with a progestin-only pill; these could include different doses, other progestin-only pills, combined oral contraceptives, or other contraceptives. Other interventions could include the timing of initiation of pills. Types of outcome measures Primary outcomes Pregnancy was the primary outcome of interest. Because of its infrequency in pill trials, we did not anticipate having sufficient power to compare pregnancy rates. However, we reported this outcome when provided. We excluded data on invalid surrogate end points (Grimes 2010) for contraceptive efficacy, specifically 3

6 ovulation. Although pregnancy cannot occur in the absence of ovulation, the presence of ovulation does not predict the risk of pregnancy, since other contraceptive mechanisms of action are involved with progestin-only pills. Secondary outcomes Side effects, including bleeding patterns, and continuation rates were secondary outcomes. Assessment of risk of bias in included studies We examined the methodological quality of trial reports according to recommended principles (Higgins 2008). We considered generation of the random sequence, allocation concealment, blinding, and losses to follow up. Adequate methods of allocation concealment included a centralized telephone system or pharmacy allocation, or use of sequentially numbered, sealed, opaque envelopes (Schulz 2002a). Excluding participants after randomization can introduce bias and threaten trial validity (Schulz 2002b). Search methods for identification of studies See: Cochrane Fertility Regulation Group methods used in reviews. Electronic searches Through October 2013, we searched for studies of progestin-only pills in MEDLINE, Cochrane Central Register of Controlled Trials (CENTRAL), POPLINE, EMBASE, and LILACS. We also searched for current trials via ClinicalTrials.gov and ICTRP. The most recent strategies can be found in Appendix 1. Previous search strategies also included EMBASE (Appendix 2; Appendix 3). Searching other resources For the initial review, we examined the reference lists of relevant articles. We also wrote to known investigators for information about other published or unpublished trials not discovered in our search. Data collection and analysis Selection of studies The first author reviewed the titles and abstracts of reports potentially eligible for inclusion. A second author confirmed the eligibility of the reports selected. Measures of treatment effect Oral contraception studies tend to have relatively high withdrawal rates. Time-to-event measures such as life-table or incidence rates (Pearl rates) are commonly used, as they are based on actual exposure to the contraceptive and prevent an imbalance in withdrawals distorting the comparisons. In this review, we attempted to extract life-table rates (actuarial or continuous) and used the rate difference as the effect measure. Where life-table rates were not published, as was common, we used the incidence rate ratio (ratio of Pearl rates) as the effect measure. We used Microsoft Excel to generate the log rate ratios and their standard errors for use in RevMan using recommended methods (Higgins 2008). Where only the crude number of events was published, we calculated the Peto odds ratio with 95% confidence interval (CI). For continuous variables, the mean difference (MD) was computed with 95% CI using a fixed-effect model. RevMan uses the inverse variance approach. Fixed effect and random effects give the same result if no heterogeneity exists, as when a comparison includes only one study. Because of disparate exposures, we were not able to combine studies in meta-analysis. Dealing with missing data We tried to contact authors for missing data and clarification of issues related to methods. For Collaborative 1998, we calculated the number of non-breast feeding women with the different bleeding patterns from the graph in the report. We were unable to do a formal comparison of rates in Sheth 1982 because standard errors were not published. R E S U L T S Data extraction and management The first author abstracted the data and entered the information into RevMan. Another author performed a second, independent data abstraction to verify the initial data entry. Any discrepancies were resolved by discussion. Description of studies Results of the search 4

7 The 2013 database searches produced 60 unduplicated references. In addition, we obtained 44 unduplicated listings from Clinical- Trials.gov and ICTRP. We did not find any new studies to include. For the initial review, six trials met our inclusion criteria. The trials enrolled 2738 participants, and sample sizes ranged from 86 to 1306 women. Trial locations included European countries, United Kingdom, United States, India, China, South Africa, Nigeria, and Kenya. Several trials were decades old and studied pills no longer commercially available; only one trial was published in the past decade. Included studies Collaborative 1998 was a multicenter trial sponsored by NV Organon and conducted in 44 centers in six countries. It compared the safety, efficacy, and acceptability of a new progestinonly pill containing desogestrel 75 µg versus a progestin-only pill containing levonorgestrel 30 µg. The trial followed 979 and 327 women using a desogestrel or levonorgestrel progestin-only pill, respectively, over 13 treatment periods of 28 days. The report provided woman-years of exposure, which we used as the denominator for pregnancy rates. Lakha 2007 compared mifepristone 5 mg versus levonorgestrel 30 µg daily for contraception. The primary outcome of interest was bleeding patterns, specifically the frequency of amenorrhea, but information on other side effects and pregnancy was gathered as well. The investigators enrolled 97 women in four cities in the UK, Hong Kong, and Africa. Seventy-four were assigned to mifepristone and 23 to the levonorgestrel progestin-only pill. Communication with the authors indicated a total of 356 and 85 cycles exposed to pregnancy, respectively. Paulsen 1974 was a report from one site of a larger randomized controlled trial sponsored by G.D. Searle and Co. It compared a progestin-only pill containing ethynodiol diacetate 0.25 mg versus a combination oral contraceptive containing ethynodiol diacetate 1.0 mg plus mestranol 0.1 mg. The trial was conducted at a Planned Parenthood clinic in California. A total of 86 women were enrolled, with 43 assigned to each treatment. Sheth 1982 reported on two centers, Bombay and Ljubljana, participating in a World Health Organization multicenter randomized trial comparing efficacy, discontinuation, and bleeding patterns of two progestin-only and two combination oral contraceptives. The four pills included norethisterone 350 µg (also known as norethindrone), levonorgestrel 30 µg, norethisterone 1 mg plus mestranol 50 µg, and levonorgestrel 150 µg plus ethinyl estradiol 30 µg. Of 599 women who agreed to participate, 518 qualified and started the assigned method; 81 women dropped out after randomization and before beginning treatment. Vessey 1972 randomized women to four different progestin-only pills in Ljubljana, Yugoslavia. These included megestrol acetate 0.7 mg, norethisterone acetate 0.3 mg, chlormadinone acetate 0.5 mg, and norgestrel mg. The investigators enrolled 450 women, with 90 to be assigned to each of five treatment groups (four progestin-only pills and one combination oral contraceptive). The planned fifth treatment group using a combined oral contraceptive was not implemented because of logistical problems. In addition, 72 (16%) of those enrolled never began their pills, and the analysis was limited to the 378 who took the drugs. Were 1997 in Eldoret, Kenya, randomized 200 women to receive norgestrel 75 µg daily beginning either six weeks postpartum versus beginning six months postpartum or when menses returned, whichever came first. Were and associates enrolled 200 women, as part of a large international trial addressing the timing of starting progestin-only pills during lactation. However, because of large losses after randomization (51% with unknown outcomes in both groups), the results cannot be considered valid. Excluded studies Two randomized controlled trials studied the new progestin-only pill containing desogestrel 75 µg (Rice 1999; Korver 2005). Both trials used ovulation as the outcome of interest, and ovulation is not a valid surrogate end point for pregnancy (Grimes 2010). We also reviewed the full text of Bhattacharya 2012 to determine the composition of the OCs studied; all were combination oral contraceptives. Risk of bias in included studies Allocation Allocation concealment was not mentioned in Collaborative 1998, and the method of randomization also was not specified. Correspondence with an author confirmed randomized permuted blocks with a block size of four, and identical-appearing blister packs were distributed sequentially. In Lakha 2007, random sequence generation was done by blocked computer-generated randomization by center, although the block lengths were not specified. Allocation concealment appears likely, since numbered pill bottles were opaque. Allocation concealment was not mentioned by Paulsen Sheth 1982 did not specify the method of random allocation. Vessey 1972 arranged pill cartons by an unspecified random process, after which serial numbers were assigned to the cartons and pill packs inside. Participants were given cartons in sequential order, which likely provided adequate concealment. Were 1997 randomly assigned participants using a computer-generated sequence developed at Family Health International; whether blocking was used is not specified in the protocol. Allocation concealment was maintained by use of sequentially numbered, opaque, sealed envelopes containing method indicator cards. 5

8 Blinding In Collaborative 1998 the two pills studied were identical in appearance. The trial was described as double blind, and correspondence confirmed that participants and investigators were kept unaware of the treatment assignment. In Lakha 2007 blinding was unlikely to have been achieved, because of dissimilar pill appearance and different bleeding patterns associated with the two treatments. Paulsen 1974 attempted to keep participants and clinicians unaware of the assigned treatment by using identical appearing pills provided by the study sponsor. The authors noted that because of the disparate bleeding patterns, clinicians could readily deduce the treatment assigned, although they felt most patients, new to oral contraception, would have less insight into expected bleeding patterns. Assessment of bleeding patterns was done by examining patient diaries blinded as to treatment. Sheth 1982 had pills re-packaged by the manufacturers in 28-day pack to blind both participants and investigators as to the treatment group. Whether the pills were identical in size, color, taste, etc., was not specified. Vessey 1972 had pills manufactured expressly for the trial; these were identical in appearance. Participants and investigators were thus unaware of the assignments to the four progestin-only pills groups. No blinding was attempted in Were 1997 comparing early versus late starting of progestin-only pills during lactation. Incomplete outcome data Collaborative 1998 excluded 14 randomized women from analysis who never began the study medications. In Lakha 2007, one randomized woman dropped out before taking her assigned drug. Overall, 57 of 74 assigned to mifepristone and 15 of 23 assigned to levonorgestrel, respectively, completed the trial. Investigators excluded six women after randomization because of persistent protocol violations. Communication with the authors provided cycles of exposure. Two pregnancies occurred in months when both mifepristone and condoms were reportedly used. In Paulsen 1974 the authors arbitrarily excluded in the reporting of bleeding analysis those participants enrolled for less than two cycles; this left 29 of 43 in the progestin-only pill group and 37 of 43 in the COC group. The trial report also noted three pregnancies during the study but did not indicate in which treatment groups these occurred. Loss to follow up was low in Sheth 1982: only 11 of 518 women (2%). Vessey 1972 reported having information for all 378 participants who took their assigned pills either until one year of follow up or until discontinuation for some reason. The life-table analysis took into account the variable lengths of follow up which are a common problem in such trials. Interpretation of Were 1997 is handicapped by high losses to follow up. In both treatment groups, 51% of participants were lost to follow up over 18 months, due in part to ethnic strife in this region of Kenya during the trial. Other potential sources of bias In Lakha 2007, the analysis excluded from consideration months in which the assigned treatment was not the sole method of contraception used. This left 356 months of exposure in the mifepristone group and 85 months in the levonorgestrel progestin-only pill group. This exclusion of months at risk of pregnancy is inappropriate; indeed, two of the pregnancies that occurred with mifepristone were in months ostensibly protected by both mifepristone and condoms. The total number of woman-months of exposure is not stated so a full analysis set is not available. Paulsen 1974 featured a simplistic, dichotomous evaluation of bleeding patterns and intentionally excluded participants after randomization; losses after randomization were high. In Vessey 1972, 16% of those enrolled never took the medication and were dropped from the study at that point. Whether these losses were different between groups is unknown. High losses after randomization occurred in Were 1997 because of civil strife; hence, these results cannot be deemed credible. Effects of interventions In the trial comparing the desogestrel versus levonorgestrel progestin-only pill (Collaborative 1998), desogestrel was not associated with a significantly lower risk of accidental pregnancy; the rate ratio was 0.27; 95% CI 0.06 to 1.19 (Analysis 1.1). Adverse events were more frequent with desogestrel (Analysis 1.2) (rate ratio 1.22; 95% CI 0.81 to 1.84, but this difference also was not statistically significant. Serious adverse events (ovarian cysts or ectopic pregnancy) were uncommon with both pills. However, the desogestrel progestin-only pill caused more bleeding problems (Analysis 1.3). Discontinuation because of irregular bleeding was more common (rate ratio 1.32; 95% CI 0.99 to 1.78). Discontinuation for all reasons was more frequent with desogestrel as well (rate ratio 1.21; 95% CI 0.99 to 1.47) (Analysis 1.4). Three pregnancies occurred amongst 979 women who started treatment with desogestrel and 4 amongst 327 who started levonorgestrel. If the reduction in risk of pregnancy was real and not due to chance, then 110 women would need to be treated with desogestrel rather than levonorgestrel to prevent one pregnancy. However, 5 more of the 110 would discontinue early because of irregular bleeding. The trial comparing low-dose mifepristone versus a levonorgestrel progestin-only pill (Lakha 2007) found a lower pregnancy rate with mifepristone (odds ratio 0.71; 95% CI 0.07 to 6.95) (Analysis 2.1). A higher prevalence of amenorrhea occurred in the former group; about half of women assigned to mifepristone had no bleeding while taking the drug (Analysis 2.2). No significant differences emerged in minor side effects between the regimens. The two treatments had different effects on the uterus. Women in the mifepristone group developed a thicker endometrium than did those assigned to the progestin-only pill (Analysis 2.3). Moreover, a higher proportion of women in the mifepristone group 6

9 required an endometrial biopsy to evaluate a dilated endometrial cavity (>12 mm) (Analysis 2.4). While no biopsy indicated serious pathology, more interventions were necessitated by mifepristone. Cystic glandular dilatation of the endometrium, a condition of undetermined medical significance, was more common in the mifepristone group. Bleeding irregularities were more common with the progestin-only pill (not currently available at this dose) than with the combined oral contraceptive in Paulsen The analysis dichotomized bleeding into regularity of cycles and intercycle bleeding. Irregular cycles occurred in all women assigned to the progestinonly pill, in contrast to those assigned to the COC (odds ratio ; 95% CI 7.63 to ) (Analysis 3.1). Intercycle bleeding was significantly more common with the progestin-only pill as well (odds ratio 6.20; 95% CI 2.11 to 18.22) (Analysis 3.2). Discontinuation from the trial was also more common with the progestin-only pill, although this difference was not statistically significant (P = 0.19) (Analysis 3.3). In the WHO four-pill comparison (Sheth 1982), pregnancy rates were highest with norethisterone 35 µg and lowest with the combination pill levonorgestrel 150 µg plus ethinyl estradiol 30 µg. At 360 days, the cumulative discontinuation rate for accidental pregnancy was lower with combined levonorgestrel and ethinyl estradiol (2.7%) than with the other pills: levonorgestrel alone (9.5%), combined norethisterone and mestranol (8.3%), and norethisterone alone (13.2%). Discontinuation because of bleeding disturbances was less common with the combination levonorgestrel and ethinyl estradiol pill than with the other three pills. On the other hand, the combination levonorgestrel and ethinyl estradiol pill had the highest rates of discontinuation because of gastrointestinal problems at 360 days. Overall discontinuation rates were similar for all four pills (Table 1). In Vessey 1972, an early small trial, accidental pregnancies were uncommon with both norethisterone acetate and norgestrel (1 in each treatment group) (Analysis 4.1). Among pills currently used, the cumulative continuation rate with norethisterone acetate at one year was significantly higher than that with norgestrel (rate difference 28.2 per 100 women, 95% CI to 44.08) (Analysis 4.2). However, discontinuation because of menstrual disturbances was significantly less common with norethisterone acetate (rate ratio 0.30; 95% CI 0.15 to 0.62) (Analysis 4.3). Discontinuation because of other side effects was also less common with norethisterone acetate, although this difference was not statistically significant (Analysis 4.4). Were 1997 found no pregnancies in either group. One group started the progestin-only pill six weeks postpartum, and the other group began the pill six months postpartum or when menses returned. Continuation rates were also similar (Table 2). By 12 months, 54% of the early-start and 57% of the late-start participants were continuing their assigned method. By 18 months, these figures were 46% and 44%, respectively. The mean duration of use was similar for both groups, as were reasons for discontinuation. D I S C U S S I O N Summary of main results Randomized controlled trials published to date are inadequate to compare progestin-only pills to each other or to combined oral contraceptives. Since progestin-only pills are commonly used during breast feeding, when fertility is low, the impact of better efficacy of any pill would be small in this setting. No trial addressed the question of consistent timing of ingestion. Likewise, a progestin-only pill may be preferred for safety reasons in older women with lower fertility who wish to continue oral contraception. Any potential benefit of better contraceptive efficacy with desogestrel 75 µg versus levonorgestrel 30 µg (Collaborative 1998) may be offset by worse bleeding patterns. These include prolonged bleeding and absence of bleeding. For every 1 pregnancy that might be prevented with desogestrel, 5 women will discontinue early because of irregular bleeding. The trade-off between efficacy and continuation may be viewed differently in different settings, and amongst women in the same settings. For some women efficacy will be the chief consideration, while for others regular bleeding will be more important. These abnormal bleeding patterns may reflect the better suppression of ovulation with the desogestrel progestin-only pill than with the levonorgestrel progestin-only pill (Rice 1999). However, the suggestion (Rice 1999) that the desogestrel pill is therefore expected to have a lower failure rate than the currently available POP [levonorgestrel 30 µg] remains unproven. While ovulation is a prerequisite for pregnancy, it does not predict the risk of pregnancy (Grimes 2010). Alternative mechanisms of progestin-only contraceptives are important, since ovulation is common with other progestin-only pills (FSRH 2008; Raymond 2011), yet pregnancy is infrequent. While daily low-dose mifepristone resulted in a higher proportion of women free from the nuisance bleeding and spotting, the likelihood of pregnancy appeared similar among women receiving mifepristone or levonorgestrel (Lakha 2007). Daily mifepristone caused endometrial thickening and distension of the uterine cavity; these required histological evaluation more often than did the progestin-only pill. To identify those at risk of hyperplasia, the investigators performed an endometrial biopsy on any participant if her uterine cavity measured >12 mm. Ovulation was better suppressed with mifepristone, but about one in five women had hormonal evidence of ovulation at any of three follow up visits. A progestin-only pill containing ethynodiol diacetate 0.25 mg caused irregular cycles in all and intercycle bleeding in most participants (Paulsen 1974). In contrast, a combined oral contraceptive containing a higher dose of the same progestin (1.0 mg) plus mestranol 0.1 mg resulted in more regular bleeding patterns. The drop-out rate was higher with the progestin-only pill than the combination oral contraceptive in this trial. 7

10 The four-group trial by the WHO (Sheth 1982) found discontinuation rates lowest with the combination levonorgestrel-ethinyl estradiol pill. The combination pill with levonorgestrel and ethinyl estradiol caused fewer bleeding problems leading to discontinuation than did the other three pills, although this was offset by more discontinuations because of gastrointestinal disturbances. The progestin-only pill containing levonorgestrel had a lower pregnancy rate than the pill with norethisterone. Poor reporting limited the usefulness of this trial. Another four-group trial (Vessey 1972) found megestrol acetate lower in efficacy than the other three pills; this difference was not statistically significant, but the sample size of only 90 per group limited the power of the study. The other notable finding was the poorer bleeding patterns seen with norgestrel in the first six months of use. The authors concluded that norethisterone acetate 30 µg appeared the most promising of the four drugs tested. Poor reporting limited the usefulness of this trial as well. The Kenyan trial of when to start a norgestrel progestin-only pill postpartum found no important differences (Were 1997). Since most women were fully or nearly fully breastfeeding during the trial, the likelihood of pregnancy was remote, and no pregnancies occurred over 18 months of observation. The high losses to follow up because of civil unrest undermined this trial. Overall completeness and applicability of evidence The age of several trials (Vessey 1972; Paulsen 1974) limited their relevance to contemporary practice, since only two products studied are currently available. Sample sizes, even in the largest trial (Collaborative 1998), were inadequate to address relative efficacy. Similarly, the comparison of levonorgestrel to mifepristone was an exploratory trial, since the latter is not used for ongoing contraception. Little information was available about the advice given as to the timing of administration of the pills or whether back up contraception should be used in case of a late or missed pill. Although accurate timing of administration is considered important for progestin-only pills, little empirical evidence supports this hypothesis, and the inter-individual variation in the metabolism of progestins is wide (Goldzieher 1994; Wallach 2000). Quality of the evidence Incomplete reporting of methods was a generic problem, including a trial (Lakha 2007) published since the CONSORT guidelines (Moher 2001; CONSORT 2010) were internationally adopted. Potential biases in the review process None evident. Agreements and disagreements with other studies or reviews A review concurred that the desogestrel pill had a lower failure rate than the levonorgestrel pill, but that the difference was not statistically significant (FSRH 2008). Collaborative 1998 lacked the power to differentiate between pregnancy rates. The review also noted that breastfeeding women may start progestin-only pills at any time after delivery, since no adverse effect of these pills on lactation has been identified (FSRH 2008). A U T H O R S C O N C L U S I O N S Implications for practice Desogestrel may be more effective than levonorgestrel, but this benefit probably applies only to women not breast feeding. In breast feeding women, the difference in efficacy was small. The potential for better efficacy with desogestrel was offset by worse bleeding patterns and more adverse events than with levonorgestrel. Given the higher cost of mifepristone and its effects on the endometrium, the prospects for daily contraception with this antiprogestin appear limited. No firm conclusion is possible concerning the comparative efficacy of progestin-only pills or whether such pills are as effective as combined oral contraceptives. One poor-quality trial suggested no large effect of timing of initiation on efficacy or continuation rates. Implications for research Trials comparing progestin-only pills versus combination oral contraceptives are needed to address comparative efficacy, an unresolved issue. Nuisance bleeding with progestin-only pills remains an obstacle to wider use. Studies of ways to reduce this problem and to support women who do experience troublesome bleeding might make this method more acceptable. A trial with sufficient power is needed to confirm the potentially higher efficacy of the desogestrel progestin-only pill. Alternatively, older progestins but with higher doses could be examined, which might allow longer half-lives and thus more time to remember the next pill. Trials might also encourage use of condoms as a backup to progestinonly pills. Future trials should report more fully (CONSORT 2010) and should provide cumulative life-table rates and their corresponding standard errors. A C K N O W L E D G E M E N T S Carol Manion of FHI 360 developed the search strategies and ran certain searches for the initial review. 8

11 R E F E R E N C E S References to studies included in this review Collaborative 1998 {published and unpublished data} Collaborative Study Group on the Desogestrel-containing Progestogen-only Pill. A double-blind study comparing the contraceptive efficacy, acceptability and safety of two progestogen-only pills containing desogestrel 75 µg/day or levonorgestrel 30 µg/day. European Journal of Contraception and Reproductive Health Care 1998;3: Lakha 2007 {published and unpublished data} Lakha F, Ho PC, Van der Spuy ZM, Dada K, Elton R, Glasier AF, et al.a novel estrogen-free oral contraceptive pill for women: multicentre, double-blind, randomized controlled trial of mifepristone and progestogen-only pill (levonorgestrel). Human Reproduction 2007;22: Paulsen 1974 {published data only} Paulsen ML, Varaday A, Brown BW Jr, Kalman SM. A randomized contraceptive trial comparing a daily progestogen with a combined oral contraceptive steroid. Contraception 1974;9: Sheth 1982 {published data only} Sheth A, Jain U, Sharma S, Adatia A, Patankar S, Andolsek L, et al.a randomized, double-blind study of two combined and two progestogen-only oral contraceptives. Contraception 1982;25: Vessey 1972 {published data only} Vessey MP, Mears E, Andolsek L, Ogrinc-Oven M. Randomised double-blind trial of four oral progestogenonly contraceptives. Lancet 1972;1: Were 1997 {published data only} Were EO, Kendall JZ, Nyongesa P. Randomised clinical trial to determine optimum initiation time of norgestrelprogestin only contraception in Eldoret Teaching Hospital, Kenya. East African Medical Journal 1997;74: References to studies excluded from this review Bhattacharya 2012 {published data only} Bhattacharya SM, Jha A. Comparative study of the therapeutic effects of oral contraceptive pills containing desogestrel, cyproterone acetate, and drospirenone in patients with polycystic ovary syndrome. Fertility and Sterility 2012;98(4): Korver 2005 {published data only} Korver T, Klipping C, Heger-Mahn D, Duijkers I, van Osta G, Dieben T. Maintenance of ovulation inhibition with the 75-µg desogestrel-only contraceptive pill (Cerazette ) after scheduled 12-h delays in tablet intake. Contraception 2005; 71:8 13. Rice 1999 {published data only} Rice CF, Killick SR, Dieben R, Coelingh Bennink H. A comparison of the inhibition of ovulation achieved by desogestrel 75µg and levonorgestrel 30 µg daily. Human Reproduction 1999;14: Additional references ACOG 2006 American College of Obstetricians and Gynecologists. ACOG technical bulletin. Use of hormonal contraception in women with coexisting medical conditions. Washington, D.C.: American College of Obstetricians and Gynecologists, Bjarnadóttir 2001 Bjarnadóttir RI, Gottfredsdóttir H, Sigurdardóttir K, Geirsson RT, Dieben TOM. Comparative study of the effects of a progestogen-only pill containing desogestrel and an intrauterine contraceptive device in lactating women. British Journal of Obstetrics and Gynaecology 2001;108: CONSORT 2010 CONSORT group. CONSORT: Transparent reporting of trials. (accessed 04 May 2011). Dunson 1993 Dunson TR, McLaurin VL, Grubb GS, Rosman AW. A multicenter clinical trial of a progestin-only contraceptive in lactating women. Contraception 1993;47: Edelman 2005 Edelman AB, Gallo MF, Jensen JT, Nichols MD, Schulz KF, Grimes DA. Continuous or extended cycle versus cyclic use of combined oral contraceptives for contraception. Cochrane Database of Systematic Reviews 2005, Issue 3. [DOI: / CD pub2] Family Health International 1991 Family Health International. Time of progestin-only oral contraceptive initiation among lactating women (protocol # 874). Internal document FFPRHC 2003 Faculty of Family Planning and Reproductive Health Care Clinical Effectiveness Unit. New product review (April 2003) desogestrel-only pill (Cerazette). Journal of Family Planning and Reproductive Health Care 2003;29: FFPRHC 2004 Faculty of Family Planning and Reproductive Health Care. FFPRHC guidance (July 2004) Contraceptive choices for breastfeeding women. Journal of Family Planning and Reproductive Health Care 2004;30:

12 FSRH 2008 Faculty of Sexual and Reproductive Healthcare Clinical Effectiveness Unit. Progestogen-only pills. (accessed 2 April 2009). Goldzieher 1994 Goldzieher JW, Fotherby K. Pharmacology of the contraceptive steroids. New York: Raven Press, Grimes 2010 Grimes DA, Schulz KF, Raymond EG. Surrogate end points in women s health research: science, protoscience, and pseudoscience. Fertility and Sterility 2010;93(6): Higgins 2008 Higgins JPT, Green S. Cochrane handbook for systematic reviews of interventions [updated February 2008]. (accessed 9 July 2008). Lidegaard 2009 Lidegaard Ø, Løkkegaard, Svendsen AL, Agger C. Hormonal contraception and risk of venous thromboembolism: national follow-up study. British Medical Journal 2009;339: b2890. Lopez 2012 Lopez LM, Newmann SJ, Grimes DA, Nanda K, Schulz KF. Immmediate start of hormonal contraceptives for contraception. Cochrane Database of Systematic Reviews 2012, Issue 12. [DOI: / CD pub3] Manchikanti 2007 Gomez AM, Grimes DA, Lopez LM, Schulz KF. Steroid hormones for contraception in women with sickle cell disease. Cochrane Database of Systematic Reviews 2007, Issue 2. [DOI: / CD pub2] McCann 1994 McCann MF, Potter LS. Progestin-only oral contraception: a comprehensive review. Contraception 1994;50:S1 S198. Moggia 1991 Mogggia AV, Harris GS, Dunson TR, Dias R, Moffia MS, Ferrer MA, et al.a comparative study of a progestinonly oral contraceptive versus non-hormonal methods in lactating women in Buenos Aires, Argentina. Contraception 1991;44: Moher 2001 Moher D, Schulz KF, Altman DG. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomised trials. Lancet 2001;357: Raymond 2011 Raymond EG. Progestin-only pills. In: Hatcher RA, Trussell J, Nelson AL, Cates W Jr, Kowal D, Policar MS editor(s). Contraceptive Technology. 20th Edition. New York: Ardent Media, Inc., 2011: Schulz 2002a Schulz KF, Grimes DA. Allocation concealment in randomised trials: defending against deciphering. Lancet 2002;359: Schulz 2002b Schulz KF, Grimes DA. Sample size slippages in randomised trials: exclusions and the lost and wayward. Lancet 2002; 359: Trussell 2011 Trussell J. Contraceptive failure in the United States. Contraception 2011;83(5): van Vliet 2011 van Vliet HA, Grimes DA, Lopez LM, Schulz KF, Helmerhorst FM. Triphasic versus monophasic oral contraceptives for contraception. Cochrane Database of Systematic Reviews 2011, Issue 11. [DOI: / CD pub3] Wallach 2000 Wallach M, Grimes DA, Chaney EJ, Connell EB, Creinin MD, Emans SJ, et al.modern oral contraception. Totowa, NJ: Enron, WHO 1998 World Health Organization collaborative study of cardiovascular disease and steroid contraception. Cardiovascular disease and use of oral and injectable progestogen-only contraceptives and combined injectable contraceptives: results of an international, multicenter, casecontrol study. Contraception 1998;57: Indicates the major publication for the study 10

13 C H A R A C T E R I S T I C S O F S T U D I E S Characteristics of included studies [ordered by study ID] Collaborative 1998 Methods Participants Interventions Outcomes Notes Stratified randomized controlled trial; double blind. Stratification: breastfeeding, switchers (use of oral contraceptives within past 2 months), or starters (not a switcher or breastfeeding). Randomized 3:1 desogestrel:levonorgestrel. Allocation concealment provided by pharmacy distribution of identical pills in identical packages 1320 healthy sexually active women aged 18 to 45 years with normal cycles; 405 women were breast feeding. Recruited at 44 centers in Europe. Mean cycle length between 24 and 35 days and intraindividual variation +/- 3 days. Willing to fill in diary card on bleeding and pill intake. Body weight between 80% and 130% of ideal. Willing to give informed consent. Exclusion criteria included contraindications to steroids, prior ectopic pregnancy, pelvic inflammatory disease, or symptomatic functional ovarian cysts Desogestrel 75 µg/day versus levonorgestrel 30 µg/day. Pregnancy, adverse events, and bleeding patterns assessed after 3, 7, and 13 treatment cycles of 28 days Both pills were identical in appearance. Sample size chosen to allow detection of a difference in incidence of 7% between the two groups for the primary bleeding pattern variables with a power of 80% and alpha of Intended sample size was 1200, in part to meet regulatory requirement for cycles of exposure. As a result of 10% overrecruitment, 1320 women were enrolled Risk of bias Bias Authors judgement Support for judgement Allocation concealment (selection bias) Low risk No mention of allocation concealment. Intentional exclusion of participants after randomization Blinding (performance bias and detection bias) All outcomes Incomplete outcome data (attrition bias) All outcomes Low risk Low risk Tablets described as identical-looking. Small numbers of participants excluded from analysis because of missing information or protocol violations Selective reporting (reporting bias) Low risk Uniform determination of outcomes for both groups. 11