Progestogen‐releasing intrauterine systems versus other forms of reversible contraceptives for contraception

Rebecca French; Annik M Sorhaindo; Huib AAM Van Vliet; Diana D Mansour; A A Robinson; Stuart Logan; Frans M Helmerhorst; John Guillebaud; Frances M Cowan

doi:10.1002/14651858.CD001776.pub2

. 2004 Jul 19;2004(3):CD001776. doi: 10.1002/14651858.CD001776.pub2

Progestogen‐releasing intrauterine systems versus other forms of reversible contraceptives for contraception

Rebecca French ¹, Annik M Sorhaindo ^2,^✉, Huib AAM Van Vliet ³, Diana D Mansour ⁴, A A Robinson ⁵, Stuart Logan ⁶, Frans M Helmerhorst ⁷, John Guillebaud ⁸, Frances M Cowan ⁹

Editor: Cochrane Fertility Regulation Group

PMCID: PMC8407482 PMID: 15266453

Abstract

Background

Hormonally impregnated intrauterine systems (IUSs) add a progestogen to a non‐medicated contraceptive device to improve contraceptive action.

Objectives

To assess the contraceptive efficacy, tolerability and acceptability of IUSs versus other reversible contraceptive methods.

Search methods

Searched databases, reference lists and relevant individuals/organisations covering the period from 1972 to July 2009.

Selection criteria

Randomised controlled trials (RCTs) comparing IUSs with other reversible contraceptives and reporting on pre‐determined outcomes, including pregnancy and continuation rates, in women of reproductive years.

Data collection and analysis

Two blinded reviewers independently assessed quality and extracted data on events per women months and single decrement life table rates for pregnancy, continuation, adverse events and reasons for discontinuation. Events per total potential number of women at follow‐up were collected for hormonal side effects and menstrual change.

Data were pooled at the same points of follow‐up to calculate rate ratios and single decrement life table rate differences. Similar interventions were combined and non‐hormonal intrauterine devices (IUDs) were divided into three categories: copper IUDs >250mm², copper IUDs ≤250mm², and non‐medicated IUDs.

Main results

Twenty‐five RCTs met the inclusion criteria and nine were included in meta‐analyses: four comparing LNG‐20 IUSs (Mirena®) with non‐hormonal IUDs, one with Norplant‐2, one with combined oral contraceptives (COCs) and three comparing P4‐IUS (Progestasert®) with non‐hormonal IUDs.

No significant difference was observed between LNG‐20 and IUD >250mm² or COC user pregnancy rates. LNG‐20 IUS users were significantly less likely to become pregnant than IUD ≤250mm² users. LNG‐20 IUS users were more likely to experience a lack of menstrual bleeding and device expulsion than IUDs >250mm² users. LNG‐20 users were significantly more likely than all IUD users to discontinue because of the lack of menstrual bleeding. They were significantly more likely to experience lack of and infrequent mentrual bleeding, but significantly less likely to experience prolonged bleeding and spotting than Norplant‐2 users.

P4‐IUS users were significantly less likely to become pregnant and more likely to discontinue than non‐medicated IUD users, but no significant difference was observed for P4‐IUS versus IUD ≤250mm² for these two outcomes. P4‐IUS users were less likely to expel the device and more likely to discontinue because of menstrual bleeding and pain than IUDs ≤250mm² users.

Authors' conclusions

Evidence suggests there is no difference in pregnancy rates among LNG‐20 IUS, IUD >250mm² and Norplant‐2 users. The LNG‐20 IUS more effectively prevented intrauterine and extrauterine pregnancies than IUDs ≤250mm². P4‐IUS was significantly more effective than non‐medicated IUDs, but no difference was observed when compared to IUDs ≤250mm². Continuation rates for LNG‐20 IUS, non‐hormonal IUDs and Norplant‐2 were similar. Lack of menstrual bleeding was the main reason for discontinuation of LNG‐20 IUS.

Recent evidence, from studies meeting the review inclusion criteria for the update conducted in July 2009, suggests that the LNG‐20 IUS does not impact upon breastfeeding performance or the growth and development of breastfed infants in lactating women nor did the device have an adverse effect on glucose metabolism among insulin‐dependent diabetic women.

Plain language summary

No difference found in pregnancy rates for women using either the LNG‐20 intrauterine system (IUS) or intra‐uterine device (IUD) for contraception

Reversible methods of contraception include the use of a system or device placed inside the uterus. The IUD is a copper device inserted into the uterus to prevent pregnancy. The intrauterine system (IUS) contains hormones that will be gradually released and provide effective contraception until removed.

The review of trials compared IUDs to IUSs and found there was no difference in the rate of unplanned pregnancies. The review found that a lack of menstrual bleeding is more likely with IUS use and that IUD use is more likely to cause heavy menstrual bleeding and pain.

Background

In the 1970s a new approach to the delivery of hormonal contraception was researched and developed. It was suggested that the addition of a progestogen to a non‐medicated contraceptive device improved its contraceptive action. An advantage of these hormonally impregnated intrauterine systems (IUS) is that they are relatively maintenance free, with users having to consciously discontinue using them to become pregnant rather than taking a proactive daily decision to avoid conception.

Levonorgestrel Intrauterine System   The levonorgestrel intrauterine system (LNG‐IUS), Mirena®, is licensed for contraceptive use in over 100 countries and is used by over 12 million women worldwide (Bayer; FDA 2000). It has a T shaped plastic frame 32 mm long with a reservoir on the vertical stem of the IUS containing 52 mg of levonorgestrel mixed with polydimethylsiloxane. This allows a steady, local release of 20µg levonorgestrel per day. Insertion of the LNG‐20 IUS may require local anaesthesia and dilatation of the cervical canal in nulliparous or peri‐menopausal woman. The net ingredient cost of the LNG‐20 IUS is more expensive than copper bearing IUDs, however it offers non‐contraceptive benefits particularly in women with heavy menstrual bleeding and may offer an alternative to hysterectomy (Bayer 2009; Hurskainen 2004; Lahteenmaki 1998).

Progesterone intrauterine system   The first IUS to be marketed was progesterone intrauterine system (P4‐IUS), Progestasert®. It has a plastic T shaped frame with a 32 mm horizontal cross bar and a 36 mm vertical stem. The vertical stem holds 38 mg of progesterone within a silicone base and when it is placed within the uterus will release 65 mcg of progesterone per day. Its contraceptive action lasts for 12‐18 months (Barnhart 1985) and is achieved by the endometrial suppression preventing implantation. A second mechanism involves the thickening of the cervical mucus preventing sperm penetration. Ovulation, however, is not affected with normal hormonal cyclical patterns demonstrated in users.

The license has not been renewed by the company in some countries in light of its reported disadvantages. These included:   ‐ yearly reinsertions with the associated risk of pelvic inflammatory disease;   ‐ increased ectopic pregnancy rate when compared to copper bearing devices;   ‐ some women experiencing persistent menstrual spotting.

Measuring contraceptive effectiveness   Extensive reviews have helped to provide greater clarity in the understanding of the various methods and terminologies employed to measure contraceptive effectiveness and have examined their relative advantages and disadvantages (Trussell 1991; Farley 1986). In brief, there are generally two methods which have been adopted, the Pearl Index (PI) and life‐tables. The PI, the older method (Pearl 1933), provides a rate per women years and is calculated by dividing the number of events (such as the number of women who discontinue using a contraceptive method) by the total number of women months and multiplying by 1200 (or 1300 if measurement is calculated by menstrual cycle). This method has been criticised because it does not account for the variation in risk of outcomes over time, nor does it account for the variation in loss to follow up (Potter 1966; Higgins 1985). Life tables do account for these factors and are therefore the most appropriate way to report contraceptive data. Confusion arises because inconsistent methods are used to define and calculate these probabilities. In brief, multiple‐decrement life table probabilities (also known as net, competing or crude rates) are calculated by working out the monthly probability of reasons for discontinuation, such as pregnancy or hormonal side effects, and multiplying these to establish the probability of discontinuation over a fixed period of time, i.e. at six months follow up, a year follow up, etc. However, single decrement life table probabilities (also known as gross, noncompeting or net rates) are recommended. They are calculated the same way but only for a single reason i.e. they censor women who discontinue a method for reasons other than the one being measured. Unfortunately, it is often impossible to distinguish which method has been used if it is not clearly stated by the authors as 'net' can be refering to single or multiple decrement probabilities.

Objectives

To determine the contraceptive effectiveness, acceptability and tolerability of IUSs. In order to do this the following questions were asked:

1. What is the relative effectiveness of IUSs in comparison to other reversible contraceptive methods?   2. What is the relative acceptability of IUSs in comparison to other reversible contraceptive methods?   3. What is the relative tolerability of IUSs in comparison to other reversible contraceptive methods?   4. What is the relative effectiveness of different types of IUS?   5. What is the relative acceptability of different types of IUS?   6. What is the relative tolerability of different types of IUS?

Methods

Criteria for considering studies for this review

Types of studies

All randomised controlled trial and controlled clinical (i.e. quasi‐randomised) trial comparisons of hormonally impregnated IUSs with other forms of reversible contraceptives.

Types of participants

Women of reproductive years

Types of interventions

Hormonally impregnated IUSs versus:   non‐hormonal IUDs   barrier contraceptives   oral contraceptives   injectable contraceptives   subdermal implants

Comparisons of different IUSs

Types of outcome measures

Primary outcome measures   Pregnancy due to method/user failure at 1, 2, 3, 4 and 5 years after starting contraceptive method   Continuation of contraceptive method after 1, 2, 3, 4 and 5 years

Secondary outcome measures   Planned pregnancy after discontinuation of contraceptive method at 1 and 2 years   Failed removal

Hormonal side effects:   Headaches   Pelvic pain   Breast tenderness   Acne   Weight gain   Nausea/vomiting   Dizziness/vertigo   Hair growth   Hair loss   Ovarian cysts   Uterine cramps   Mood changes   Loss of libido

Menstrual bleeding changes (using termionology recommended by Fraser 2007):   Painful menstruation   Spotting   Infrequent menstrual bleeding   Lack of menstrual bleeding   Heavy menstrual bleeding   Prolonged bleeding   Irregular bleeding

Local device problems:   Malposition   Translocation   Expulsion

Adverse clinical events:   Ectopic pregnancy   Pelvic inflammatory disease   Sexually transmitted infections   Anaemia   Breast cancer   Fibroids   Vaginitis   Urinary tract infection   Cervical intraepithelial neoplasia I   Cervical intraepithelial neoplasia II   Cervical intraepithelial neoplasia III   Invasive cervical cancer   Myocardial infarction   Stroke   Pulmonary embolism/thrombophlebitis   Gall bladder disease   Death

Reason for discontinuation:   Hormonal side effects   Menstrual change   Adverse clinical event   Local device problem   Planning pregnancy   Patient choice ‐ other

Search methods for identification of studies

We obtained relevant randomized and controlled clinical trials from a search of publications describing IUSs. We conducted computerized searches of the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, POPLINE, LILACS and Web of Science using the following search strategy:

CENTRAL   intrauterine devices medicated

MEDLINE   intrauterine devices, medicated   AND   (clinical trial*)

EMBASE   1: intrauterine devices, medicated   2: intrauterine contraceptive device   3: IUS   4: 1 or 2 or 3   5: norgestrel   6: levonorgestrel   7: keto(w)desogestrel   8: etonorgestrel   9: P4‐IUS or P4‐IUS intrauterine progesterone contrac*   10: Mirena or Mirena coil or Mirena IUS   11: levonova   12: 5 or 6 or 7 or 8 or 9 or 10 or 11   13: 4 and 12

POPLINE   (kw) iud hormone releasing   AND   (subject or textword) clinical trial*

LILACS   intrauterine devices, medicated

In a previous version of this review the Science Citation Index and Psych. Lit. databases were searched from 1972 to 1998 July using the strategy:

#1 "INTRAUTERINE‐DEVICES,‐MEDICATED" / all subheadings   #2 INTRAUTERINE SYSTEM* or IUS*   #3 explode "NORGESTREL" / all subheadings   #4 "LEVONORGESTREL"/all subheadings   #5 NORGESTREL   #6 LEVONORGESTREL   #7 KETO near DESOGESTREL   #8 ETONORGESTREL   #9 P4‐IUS   #10 MIRENA   #11 LEVONOVA   #12 #1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11

WEB OF SCIENCE (November 2003 ‐ July 2009)

The reference lists of all identified publications were searched for previously unidentified articles.

We contacted the relevant pharmaceutical companies and asked to release results of any relevant unpublished studies for inclusion in the review. Individuals and organisations with an interest in IUS research and databases housing information on clinical trials were solicited to identify unpublished and ongoing studies relevant to the review.

Data collection and analysis

The selection of studies for inclusion and their methodological quality were independently assessed and reported by reviewers (RF, AS, FC and HV). Quality assessment forms were designed, and included general methodological factors, as well as some of contraceptive specific factors recommended by Trussell 1991.The following quality factors were included on the checklist:   ‐ method of randomization described,   ‐ allocation concealment,   ‐ blinded assessment of outcomes,   ‐ groups treated identically other than named intervention,   ‐ description of women who withdrew or were lost to follow up provided,   ‐ description of hormonal contraceptive method or pregnancy immediately prior to study enrolment,   ‐ statistical method (with reference) used to analyse pregnancy and continuation of methods,   ‐ description of contraceptive failure provided (i.e. user or method failure or both),   ‐ active follow up conducted (i.e. analysis of follow up delayed a few months to allow inclusion of undetected pregnancies)

Contraceptive effectiveness and continuation

Single‐decrement life table probabilities with their standard errors (SEs), and events per women months, akin to the Pearl Index rate, were collected for each outcome at specific follow up points (at one, two, three, four and five years). It was decided to collect both ways of reporting event rates as, although single‐decrement rates are the ideal, they are not commonly employed and there was usually sufficient information in the papers to collect events per women months. Of those papers which had reported single decrement probabilities, only a few had given SEs, a necessity for meta‐analysis. Authors who had used single decrement probabilities but had not given their SEs were contacted and asked to provide them where possible. Unless otherwise stated, in the rest of the text life table probabilities refers to single decrement life tables for any discontinuation outcomes.

In order to obtain a relative measure of continuation taking account of the time the method was used, the number of women months contributing to follow up and the number of potential women months at the specified time points were collected. Potential women months were calculated by multiplying the number of women recruited onto each of the studies with the total number of months at each of the specified time points (e.g. at one year the number of women recruited into a study was multiplied by 12 months). This method has been described as a way of measuring completeness to follow‐up (Clark 2002).

Menstrual changes, hormonal side‐effects and adverse events

Menstrual change outcomes were only collected if investigators had stipulated that they had been measured over 90 day intervals as recommended by Rodriguez 1976. Lack of menstrual bleeding was no bleeding or spotting (B‐S) throughout the reference period (RP). Infrequent bleeding was less than three B‐S episodes starting within a RP excluding no menstrual bleeding; frequent bleeding was more than five B‐S episodes starting within a RP; and prolonged bleeding was at least one B‐S episode lasting greater than 14 days starting within a RP. The number of events and total number of women at each 90 day interval were collected to calculate risk ratios for menstrual change outcomes.

Data on hormonal side effects and planned pregnancy (after discontinuation of contraceptive method) were collected at yearly time intervals. Data on these outcomes were only collected if the investigators provided number of events and total number of women at follow up, so that risk ratios for each of the side effects identified in the protocol could be determined. Data on weight change were collected by extracting the mean weight difference, with its standard deviation, between the contraceptive methods under investigation.

Data systhesis

A description of the demographic characteristics of the study participants, the interventions, environmental and geographical factors which may influence findings, quality and the measured outcomes were collected, so that a decision could be made about the results of individual studies and whether it was feasible to combine the data.

Studies were only combined when the comparative interventions were similar, such as IUSs versus subdermal implants or IUSs versus non‐hormonal IUDs contraceptives. Non‐hormonal IUDs were divided into three categories for the purpose of data synthesis. The first, defined as IUDs >250mm2, included CuT 380A and CuT 380Ag IUDs; the second, defined as IUDs <=250mm2, included the Nova‐T, Multiload, CuT 200 and CuT 220 IUDs; and the third were non‐medicated IUDs. The first two categories were based on the surface area of the copper wire. In situations where it was not possible or appropriate to synthesise data, a narrative description is provided.

In order to obtain a summary effect size of an event per women months the rate ratios of the treatment and comparison events were combined. This method gave a relative measure of 'treatment' effect, that is how much more or less likely IUS users experienced an event in comparison to users of other contraceptive methods. The log rate ratios and their variances for events were calculated for each study (Hasselblad 1995). It was then possible to calculate the inverse weighted average of the log rate ratios. Events were only combined if they were measured over the same time period (i.e. one year, two years and so on) because of their variability over time. For the purpose of data synthesis, in situations where there were no events in one arm of the trial a continuity correction was implemented by adding a half to each cell.

In order to synthesise life table probabilities, it was necessary to calculate the absolute measurement of 'treatment' effect. This was done by subtracting the comparison group probability from the intervention group probability. The SE for the measurement of true effect was then calculated by obtaining the square root of sum of the squared SE of the intervention group probability and the squared SE of the comparison group probability. If there was a probability of zero in one of the groups, its SE was assumed to be the same as the SE of the probability in the comparison group. The inverse weighted average of the rate differences was then calculated. It was thus possible to obtain an absolute difference in percentage terms of 'treatment' effect, that is the attributable risk, between IUS users and users of other contraceptive methods.

To order to obtain pooled estimates for risk ratios and mean differences, the inverse variance weighted average was used with the sample log risk ratio and the sample mean difference, respectively, calculated from each study (Petitti 1994). A continuity correction was performed when necessary as described above for the calculated rate ratios.

Microsoft Excel was used to calculate the pooled effect sizes as it was not possible to calculate rate ratios or life table differences in RevMan.

The degree of heterogeneity was investigated and reported. A random effects approach was used for the meta‐analysis (Dersimonian 1986). In the absence of heterogeneity this coincides with a fixed effect analysis. No statistical heterogeneity was identified in the analyses unless explicitly stated in the results below.

An economic evaluation was conducted using the results of the systematic review and meta‐analysis, and this has been published elsewhere (French 2000)

Results

Description of studies

Twenty‐five RCTs comparing hormonally impregnated IUSs to a reversible contraceptive method met the inclusion criteria (See Included studies). Eleven trials were conducted in developing or transitional countries (Affandi 1980; WHO 1988; Baveja 1989; El Mahgoub 1982; Kapur 2008; Lavin 1983; Piazarro 1977; Rogovskaya 2005; Shaamash 2005; Wang 1992; Zhu 1991) eight in developed countries (Andersson 1994; Fylling 1979; Heikkila 1982; Janssen 2000; Larsen 1981; Pakarinen 1996; Rybo 1983; Suhonen 2004) and five were international multicentre studies conducted in both developed and developing countries (Luukkainen 1986; Sivin 1994;WHO 1983; WHO 1987; WHO 1988). In one publication it was not possible to determine the study setting (Newton 1979). The majority of trials (12) were set in community‐based family planning clinics.

The age range of participants varied from 15 ‐ 45 years. None of the studies confined entry to specific age requirements, other than ensuring the recruited women were of reproductive age. Seventeen of the 25 trials limited recruitment to women with proven fertility (Andersson 1994; WHO 1988; Baveja 1989; Heikkila 1982; Kapur 2008; Lavin 1983; Luukkainen 1986; Piazarro 1977;Rogovskaya 2005; Rybo 1983; Shaamash 2005; Sivin 1994; WHO 1987; Wang 1992; El Mahgoub 1982; Zhu 1991). Four studies recruited women post partum or post abortion (Heikkila 1982; Lavin 1983; El Mahgoub 1982; Shaamash 2005). Two studies restricted recruitment to women who were breast feeding (Heikkila 1982; Shaamash 2005). Five studies stated that they only included women with regular menstrual cycles (Baveja 1989; Janssen 2000; Pakarinen 1996; Piazarro 1977; Zhu 1991).

Nearly all of the interventions were either comparisons of IUSs with different hormonal release rates or of IUSs versus non‐hormonal IUDs. There were two exceptions: a comparison of LNG‐20 IUS with Norplant‐2 (Wang 1992) and a comparison with combined oral contraceptives (Suhonen 2004).

It was documented in two of the 21 trials that contraceptive counselling had been provided (Andersson 1994; Wang 1992). None of the studies mentioned any specific training for those inserting the devices.

Risk of bias in included studies

Details of the methodological quality of each of the studies are provided in the Characteristics of included studies. It was documented that allocation of contraceptive method was concealed to the investigator in eleven trials (Andersson 1994; Baveja 1989; Kapur 2008; Newton 1979; Pakarinen 1996; Rogovskaya 2005; Shaamash 2005; Sivin 1994; Wang 1992; WHO 1983). It was reported that investigators were blind to contraceptive method when assessing outcomes in only four of the trials (Janssen 2000; Luukkainen 1986; Newton 1979; Piazarro 1977). Women were blind to allocated method in an additional four studies (Andersson 1994; Janssen 2000; Larsen 1981; Rogovskaya 2005).

In 18 studies, the compared groups were treated identically in terms of measurement of outcomes (Andersson 1994; Baveja 1989; Fylling 1979;Kapur 2008 Janssen 2000; Larsen 1981; Lavin 1983; Luukkainen 1986; Newton 1979; Pakarinen 1996; Piazarro 1977; Rybo 1983; Shaamash 2005; Sivin 1994; Suhonen 2004; Wang 1992; WHO 1983; WHO 1987). A description of the characteristics of women lost to follow up or who withdrew from the study was not provided in any of the publications.

Twelve studies used life table analysis to determine pregnancy and continuation rates (Andersson 1994; Baveja 1989; El Mahgoub 1982; Larsen 1981; Luukkainen 1986; Newton 1979; Pakarinen 1996; Piazarro 1977; Sivin 1994; Wang 1992; WHO 1983; WHO 1987). It was possible to determine whether single or multiple decrement probabilities had been reported in nine of these studies (Andersson 1994; Baveja 1989; Larsen 1981; Luukkainen 1986; Pakarinen 1996; Sivin 1994; Wang 1992; WHO 1983; WHO 1987) and all but one provided single decrement probabilities (Larsen 1981).

Less than half of all studies provided information of contraceptive methods used or pregnancy immediately prior to enrolment (Andersson 1994; WHO 1988; El Mahgoub 1982; Heikkila 1982; Lavin 1983; Luukkainen 1986; Piazarro 1977; Wang 1992). In the 15 studies where pregnancy occurred, nine distinguished between user or method failure (or both) (Andersson 1994; Baveja 1989; Luukkainen 1986; Pakarinen 1996; Piazarro 1977; Sivin 1994; Wang 1992; WHO 1983; WHO 1987). Active follow up was conducted in three trials (Sivin 1994; WHO 1983; WHO 1987).

Effects of interventions

Some studies which would have met the inclusion criteria but examined prototype contraceptive methods or methods that are not (longer) available were excluded from the meta‐analyses (El Mahgoub 1982; Heikkila 1982; Janssen 2000; Pakarinen 1996; WHO 1983; WHO 1987).

LNG‐20 versus non‐hormonal IUD >250mm2

Five studies compared the LNG‐20 IUS with the non‐hormonal IUD >250mm2 (Baveja 1989; Kapur 2008; Rogovskaya 2005; Shaamash 2005; Sivin 1994). It was possible to extract data from two of these studies for the meta analysis (Baveja 1989; Sivin 1994). Rate ratios and single decrement life table differences derived from the two studies are presented in Table 1 and Table 2, respectively (for the following outcomes: pregnancy, continuation, expulsion, embedded device, ectopic pregnancy, PID, and discontinuation due to hormonal side effects, menstrual side effects, adverse events, planning a pregnancy and/or personal choice). The relative risk for planned pregnancy after removal of the LNG‐20 IUS compared to CuT 380 Ag IUD was 1.25 (95% CI 0.45 to 3.48) at one year (Sivin 1994) Figure 1. It was possible to extract data on menstrual change outcomes from one study only (Sivin 1994). Data from this study indicated that women using LNG‐20 IUSs were more likely to experience no menstrual bleeding than women using CuT 380Ag IUDs and this risk increased over time, at three months the risk ratio was 2.35 (95% CI 1.37 to 4.04) Figure 2 which increased to 11.08 (95% CI 6.61 to 18.57) at three years follow up. No significant differences were noticed between LNG‐20 IUS and CuT 380Ag IUDs in terms of prolonged bleeding, with risk ratios of 0.88 (95% CI 0.55 to1.39) at three months and 0.15 (95% CI 0.02 to1.10) at three years Figure 3. It was not possible to extract data for any other menstrual change outcomes, but Sivin et al (1994) reported that LNG‐20 IUS users were significantly less likely to experience painful menstruation.

1. LNG‐20 versus IUD >250mm2: Rate ratios with 95% CI.

Outcome	One year	Two years	Three years	Four years	Five years
Pregnancy	1.01 (0.71 ‐ 5.82)   [Sivin 1994]	0.30 (0.07 ‐ 1.24)   [Sivin 1994]	0.11 (0.01 ‐ 2.12)   [Baveja 1989]	Not available	0.66 (0.25 ‐ 1.75)   [Sivin 1994]
Continuation	0.97 (0.90 ‐ 1.06)   [Sivin 1994, Baveja 1989]	0.94 (0.86 ‐ 1.04)   [Sivin 1994, Baveja 1989]	0.89 (0.71 ‐ 1.11)   [Baveja 1989]	Not available	0.91 (0.78 ‐ 1.06)   [Sivin 1994]
Expulsion	1.11 (0.72 ‐ 1.71)   [Sivin 1994]	Not available	Not available	Not available	1.53 (1.13 ‐ 2.07)   [Sivin 1994]
Embedded	Not available	Not available	Not available	Not available	7.0 (0.36 ‐ 135.52)   [Sivin 1994]
Ectopic   Pregnancy	None	None	Not available	Not available	0.22 (0.01 ‐ 4.56)   [Sivin 1994]
PID	1.23 (0.50 ‐ 3.03)   [Sivin 1994]	Not available	Not available	Not available	Not available
Discontinuation:   Hormonal	0.81 (0.23 ‐ 2.80)   [Sivin 1994]	Not available	1.71 (0.64 ‐ 4.55)   [Baveja 1989]	Not available	4.24 (1.99 to 9.05)   [Sivin 1994]
Discontinuation:   All Menstrual	1.48 (1.02 ‐ 2.14)   [Sivin 1994]	Not available	Not available	Not available	1.48 (1.23 ‐ 1.79)   [Sivin 1994]
Discontinuation:   Menstrual ‐ Bleeding   & pain	Not available	Not available	Not available	Not available	0.71 (0.56 ‐ 0.89)   [Sivin 1990]
Discontinuation:   Menstrual ‐ Pain only	0.80 (0.41 ‐ 1.56)   [Sivin 1994]	Not available	Not available	Not available	Not available
Discontinuation:   Menstrual: Absence of menstrual bleeding	65.51 (4.01 ‐ 1069.85)   [Sivin 1994]	Not available	Not available	Not available	48.92 (16.93 ‐ 141.36)   [Sivin 1994]
Discontinuation:   Adverse event	Not available	Not available	1.03 (0.18 ‐ 5.92)   [Baveja 1989]	Not available	Not available
Discontinuation:   Planning pregnancy	0.94 (0.47 ‐ 1.89)   [Sivin 1994]	Not available	Not available	Not available	1.11 (0.89 ‐ 1.39)   [Sivin 1994]

Outcomes	One year	Two years	Three years	Four years	Five years
Pregnancy	‐0.16 (‐0.65 to 0.34)   [Sivin 1994, Baveja 1989]	‐1 (‐2.39 to 0.39)   [Baveja 1989]	‐1 (‐2.39 to 0.39)   [Baveja 1989]	Not available	‐0.3 (‐1.56 to 0.96)   [Sivin 1994]
Continuation	‐6.3 (‐10.00 to ‐2.56)   [Sivin 1994]	‐8.1 (‐12.40 to ‐3.80)   [Sivin 1994]	Not available	Not available	‐7.6 (‐11.90 to ‐3.30)   [Sivin 1994]
Expulsion	0.84 (‐1.19 to 2.88)   [Sivin 1994, Baveja 1989]	2.1 (‐1.64 to 5.84)   [Baveja 1989]	3 (‐1.17 to 7.17)   [Baveja 1989]	Not available	4.4 (1.46 to 7.34)   [Sivin 1994]
PID	0.3 (‐0.96 to 1.56)   [Sivin 1994]	Not available	Not available	Not available	Not available
Discontinuation:   Hormonal	‐0.1 (‐1.21 to 1.01)   [Sivin 1994]	Not available	Not available	Not available	Not available
Discontinuation:   All menstrual	6.91 (2.87 to 10.94)   [Sivin 1994, Baveja 1989]	11.1 (6.26 to 15.94)   [Baveja 1989]	14.5 (8.78 to 20.22)   [Baveja 1989]	Not available	Not available
Discontinuation:   Menstrual ‐ Bleeding   & pain	Not available	Not available	Not available	Not available	‐7.9 (‐10.89 to ‐4.91)   [Sivin 1994]
Discontinuation:   Mentrual ‐ Pain only	‐0.9 (‐2.86 to 1.06)	Not available	Not available	Not available	Not available
Discontinuation:   Menstrual ‐ Absence of menstrual bleeding	5.04 (3.19 to 6.90)   [Sivin 1994, Baveja 1989]	9.5 (6.27 to 12.73)   [Baveja 1989]	13.3 (9.30 to 17.30)   [Baveja 1989]	Not available	19.3 (16.14 to 22.46)   [Sivin 1994]
Discontinuation: Planning pregnancy	‐0.1 (‐2.04 to 1.84)   [Sivin 1994]	Not available	Not available	Not available	1.5 (‐3.50 to 6.50)   [Sivin 1994]
Discontinuation: Personal choice	0.8 (‐1.01 to 2.61)   [Sivin 1994]	Not available	Not available	Not available	0.1 (‐3.50 to 3.70)   [Sivin 1994]

Outcome	One year
Discontinuation: Hormonal	1.00 (0.32 ‐ 3.07) [Suhonen 2004]
Discontinuation: Planning pregnancy	0.21 (0.01 ‐ 4.39) [Suhonen 2004]
Discontinuation: Patient choice	1.40 (0.48 ‐ 4.02) [Suhonen 2004]

Date	Event	Description
15 April 2008	Amended	Converted to new review format.
24 May 2004	New citation required and conclusions have changed	Substantive amendment

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pregnancy due to method failure			Other data	No numeric data
1.1 At 1 year			Other data	No numeric data
1.2 At 2 years			Other data	No numeric data
1.3 At 3 years			Other data	No numeric data
1.4 At 5 years			Other data	No numeric data
2 Continuation of method			Other data	No numeric data
2.1 At 1 year			Other data	No numeric data
2.2 At 2 years			Other data	No numeric data
2.3 At 3 years			Other data	No numeric data
2.4 At 5 years			Other data	No numeric data
3 Planned pregnancy after discontinuation of method	1	86	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.25 [0.45, 3.48]
3.1 At 1 year	1	86	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.25 [0.45, 3.48]
4 Amenorrhoea	1	700	Peto Odds Ratio (Peto, Fixed, 95% CI)	5.29 [3.64, 7.68]
4.1 At 3 months	1	441	Peto Odds Ratio (Peto, Fixed, 95% CI)	2.35 [1.37, 4.04]
4.2 At 3 years	1	259	Peto Odds Ratio (Peto, Fixed, 95% CI)	11.08 [6.61, 18.57]
5 Prolonged bleeding	1	700	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.80 [0.51, 1.26]
5.1 At 3 months	1	441	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.88 [0.55, 1.39]
5.2 At 3 years	1	259	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.15 [0.02, 1.10]
6 Expulsion			Other data	No numeric data
6.1 At 1 year			Other data	No numeric data
6.2 At 2 years			Other data	No numeric data
6.3 At 3 years			Other data	No numeric data
6.4 At 5 years			Other data	No numeric data
7 Embedded			Other data	No numeric data
7.1 At 5 years			Other data	No numeric data
8 Ectopic pregnancy			Other data	No numeric data
8.1 At 1 year			Other data	No numeric data
8.2 At 2 years			Other data	No numeric data
8.3 At 5 years			Other data	No numeric data
9 Pelvic inflammatory disease			Other data	No numeric data
9.1 At 1 year			Other data	No numeric data
10 Hormonal reasons for discontinuation			Other data	No numeric data
10.1 At 1 year			Other data	No numeric data
10.2 At 3 years			Other data	No numeric data
10.3 At 5 years			Other data	No numeric data
11 Menstrual reasons for discontinuation: all			Other data	No numeric data
11.1 At 1 year			Other data	No numeric data
11.2 At 2 years			Other data	No numeric data
11.3 At 3 years			Other data	No numeric data
11.4 At 5 years			Other data	No numeric data
12 Menstrual reasons for discontinuation: bleeding & pain			Other data	No numeric data
12.1 At 5 years			Other data	No numeric data
13 Menstrual reasons for discontinuation: pain			Other data	No numeric data
13.1 At 1 year			Other data	No numeric data
13.2 At 5 years			Other data	No numeric data
14 Menstrual reasons for discontinuation: absence of menstrual bleeding			Other data	No numeric data
14.1 At 1 year			Other data	No numeric data
14.2 At 5 years			Other data	No numeric data
15 Discontinuation due to adverse event			Other data	No numeric data
15.1 At 3 years			Other data	No numeric data
16 Discontinuation because planning pregnancy			Other data	No numeric data
16.1 At 1 year			Other data	No numeric data
16.2 At 5 years			Other data	No numeric data
17 Personal reasons for discontinuation			Other data	No numeric data
17.1 At 1 year			Other data	No numeric data
17.2 At 5 years			Other data	No numeric data

Pregnancy due to method failure
Study
At 1 year
Baveja 1989	Single decrement life table probabilities (SE) = 0.0 (0.4) vs. 0.8 (0.4)
Sivin 1994	2/7680 women months vs. 2/7740 women months   Single decrement life table probabilities (SE) = 0.3 (0.2) vs. 0.3 (0.2)
At 2 years
Baveja 1989	Single decrement life table probabilities (SE) = 0.0 (0.5) vs. 1.0 (0.5)
Sivin 1994	2/19644 women months vs. 7/20436 women months
At 3 years
Baveja 1989	0/10589 women months vs. 4/10869 women months   Single decrement life table probabilities (SE) = 0.0 (0.5) vs. 1.0 (0.5)
At 5 years
Sivin 1994	6/34944 women months vs. 10/38268 women months   Single decrement life table probabilities (SE) = 1.1 (0.5) vs. 1.4 (0.4)

Pregnancy due to method failure
Study
At 1 year
Andersson 1994	1/18664 women months vs. 8/9326 women months
Baveja 1989	Single decrement life table probabilities (SE) = 0.0 vs. CuT 220C 0.0 and vs. CuT 200B 0.9 (0.4)
Luukkainen 1986	1/1654 women months vs. 4/1708 women months
At 2 years
Baveja 1989	Single decrement life table probabilities (SE) = 0.0 vs. CuT 220C 0.0 and vs. CuT 200B 0.9 (0.4)
At 3 years
Andersson 1994	3/46200 women months vs. 24/23568 women months
Baveja 1989	0/10589 women months vs. 7/24225 women months (vs. CuT 220C 1/12076 women months and vs. CuT 220B 6/12149 women months)   Single decrement life table probabilities (SE) = 0.0 vs. CuT 220C 0.3 (0.3) and vs. CuT 200B 1.6 (0.6)
At 5 years
Andersson 1994	5/67380 women months vs. 35/33312 women months
Luukkainen 1986	1/5495 women months vs. 7/5176 women months

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Pregnancy			Other data	No numeric data
1.1 At 1 year			Other data	No numeric data
1.2 At 2 years			Other data	No numeric data
1.3 At 3 years			Other data	No numeric data
2 Continuation of method			Other data	No numeric data
2.1 At 1 year			Other data	No numeric data
3 Expulsion			Other data	No numeric data
3.1 At 1 year			Other data	No numeric data
4 Breast cancer			Other data	No numeric data
4.1 At 1 year			Other data	No numeric data
5 Ovarian cysts			Other data	No numeric data
5.1 At 1 year			Other data	No numeric data
6 Spotting	1		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
6.1 At 1 year	1	186	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.26 [0.14, 0.51]
6.2 At 2 years	1	158	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.19 [0.08, 0.45]
6.3 At 3 years	1	134	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.20 [0.08, 0.50]
7 Infrequent menstrual bleeding	1		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
7.1 At 1 year	1	186	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.77 [0.93, 3.37]
7.2 At 2 years	1	158	Peto Odds Ratio (Peto, Fixed, 95% CI)	7.16 [3.56, 14.40]
7.3 At 3 years	1	134	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.07 [0.38, 3.03]
8 Absence of menstrual bleeding	1		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
8.1 At 1 year	1	186	Peto Odds Ratio (Peto, Fixed, 95% CI)	2.47 [1.06, 5.72]
8.2 At 2 years	1	158	Peto Odds Ratio (Peto, Fixed, 95% CI)	9.89 [3.96, 24.72]
8.3 At 3 years	1	134	Peto Odds Ratio (Peto, Fixed, 95% CI)	2.61 [0.57, 11.92]
9 Prolonged bleeding	1		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
9.1 At 1 year	1	186	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.15 [0.08, 0.32]
9.2 At 2 years	1	158	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.18 [0.08, 0.40]
9.3 At 3 years	1	134	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.20 [0.07, 0.56]

Pregnancy
Study
At 1 year
Wang 1992	1/1157 women months vs. 0/1187 women months
At 2 years
Wang 1992	1/2171 women months vs. 0/2218 women months
At 3 years
Wang 1992	1/3098 women months vs. 0/3093 women months

Pregnancy
Study
At 1 year
Suhonen 2004	0/1128 women months vs. 0/1188 women months

Pregnancy
Study
At 1 year
Newton 1979	3/3389 women months vs. 28/2953 women months

Outcome	One year
Pregnancy	1.41 (0.57 ‐ 3.51)   [Larsen 1981; Fylling 1979]
Continuation	0.97 (0.78 ‐ 1.23)   [Larsen 1981]
Expulsion	0.11 (0.03 ‐ 0.43)   [Fylling 1979]
Ectopic pregnancy	3.57 (0.39 ‐ 32.36)   [Larsen 1981;Fylling 1979]

Outcome	One year
Pregnancy	0.09 (0.03 ‐ 0.28)   [Newton 1979]
Expulsion	0.95 (0.54 ‐ 1.66)   [Newton 1979]
Ectopic pregnancy	0.29 ‐ 7.13)   [Newton 1979]
Discontinuation:   Planning pregnancy	1.29 (0.88 ‐ 1.90)   [Newton 1979]
Discontinuation:   Personal reasons	0.46 (0.20 ‐ 1.07)   [Newton 1979]

Pregnancy
Study
At 1 year
Fylling 1979	7/1729 women months vs. 3/1483 women months
Larsen 1981	4/1996 women months vs. 4/1943 women months

Methods	Setting: Indonesia   697 women randomised   Follow up: 2 years
Participants	Not stated
Interventions	P4‐IUS [n=72] vs. CuT 200, Cu 7 and Lippes loop IUDs [n=75, 75 and 75, respectively]
Outcomes	Pregnancy   Reasons for discontinuation
Notes	Abstract Quality assessment not conducted
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	Setting: Multinational (Denmark, Finland, Hungary, Norway and Sweden), Family Planning Clinics (12)   2758 women randomised   Follow up: 5 years
Participants	18‐38 years   Parous   Not breast feeding
Interventions	LNG‐20 IUS [n=1821] vs. Nova‐T IUD [n=937]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation   Adverse events   Hormonal side effects   Pregnancy after discontinuation of method
Notes	Quality assessment:   Randomisation technique: No mention   Allocation concealment technique: Centrally prepared envelopes   Description of prior contraceptive method / pregnancy provided   Mesurement: Groups treated identically   Method of analysis: Life tables (multiple and single decrement rates)   User/method failure reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	Setting: India, Family Planning Clinics   2118 women randomised   Follow up: 3 years
Participants	18‐40 years   Proven fertility   Regular menses
Interventions	LNG‐20 IUS [n=475] vs. CuT 380Ag, CuT 220C and CuT 200B IUDs [n=434, 496 and 500, respectively]
Outcomes	Pregancy   Continuation   Reasons for discontinuation   Menstrual change
Notes	Quality assessment:   Randomisation technique: Computed random numbers   Allocation concealment technique: Sealed envelopes   Measurement: Groups treated identically   Method of analysis: Life tables (single decrement rates)   User / method failure reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Methods	Setting: Egypt, Family Planning Clinics   300 women randomised   Follow up: 3 years
Participants	15‐40 years   Parous   Hormonal contraceptive users at enrolment and immediate post partum women excluded
Interventions	LNG‐10 IUS and Norgestrel T (various doses) IUSs vs. CuT 200 IUD [n=100 in each group]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation   Menstrual change and blood loss   Endometrial and cervical cell changes
Notes	Quality assessment:   Randomisation technique: No mention   Allocation concealment technique: No mention   Description of prior contraceptive method / pregnancy provided   Method of analysis: Life tables (method not stated)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	Setting: Denmark   326 women randomised   Follow up: 1 year
Participants	Mixed parity
Interventions	P4‐IUS [n=162] vs. Nova‐T IUD [n=164]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation   Serum immunoglobin levels
Notes	Quality assessment:   Randomisation technique: No mention   Allocation concealment technique: No mention   Measurement: Groups treated identically   Method of analysis: Other
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	Setting: Finland, Maternity Unit   80 women randomised   Follow up: 1 year
Participants	Postpartum   Amenorrhoeic   Breast feeding
Interventions	LNG‐30 IUS[n=40] vs. Nova‐T IUD [n=40]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation   Hormonal side effects   Menstrual change   LNG plasma concentration
Notes	Quality assessment:   Randomisation technique: No mention   Allocation concealment technique: No mention   Description of prior contraceptive method / pregnancy provided   Method of analysis: Other   User / method failure reported: Not applicable
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	Setting: The Netherlands, University Medical Centre   203 women randomised   Follow up: 2 years
Participants	18‐45 years   Variable parity   Regular menses
Interventions	Multiload Cu250 releasing different doses of 3‐keto‐ desogestrel [n= 151] vs. Multiload Cu250 [n=51]
Outcomes	Menstrual blood loss   Hemoglobin, ferritin and 3‐keto‐desogestrel concentration   Bleeding pattern   Subjective complaints
Notes	Quality assessment: Randomisation technique: Computer generated randomization list.   Allocation concealment technique: No mention.   Double‐blinded assessment of outcomes.   Description of prior contraceptive method / pregnancy not provided   Measurement: Groups treated identically   Method of analysis: Other
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	Setting: India 170 women randomised Follow‐up: 1 year
Participants	26‐35 years Parous, desiring contraception
Interventions	LNG IUS [n=70] vs. Cu T 380 [n=70]
Outcomes	Pregnancy Menstrual change Reasons for discontinuation
Notes	Quality assessment:   Randomisation technique: no mention   Allocation concealment technique: no mention   Women blinded to method   Measurement: Groups treated identically   Method of analysis: Other
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	Setting: Denmark   382 women randomised   Follow up: 1 year
Participants	15‐44 years   Variable parity
Interventions	P4‐IUS [n=196] vs. CuT 200 IUD [n=186]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation
Notes	Quality assessment:   Randomisation technique: No mention   Allocation concealment technique: No mention   Women blinded to method   Measurement: Groups treated identically   Method of analysis: Life tables (multiple decrement rates)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

PERMALINK

Progestogen‐releasing intrauterine systems versus other forms of reversible contraceptives for contraception

Rebecca French

Annik M Sorhaindo

Huib AAM Van Vliet

Diana D Mansour

A A Robinson

Stuart Logan

Frans M Helmerhorst

John Guillebaud

Frances M Cowan

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Search methods for identification of studies

Data collection and analysis

Contraceptive effectiveness and continuation

Menstrual changes, hormonal side‐effects and adverse events

Data systhesis

Results

Description of studies

Risk of bias in included studies

Effects of interventions

LNG‐20 versus non‐hormonal IUD >250mm2

1. LNG‐20 versus IUD >250mm2: Rate ratios with 95% CI.

2. LNG‐20 IUS versus IUD >250mm2: Life table differences with 95%CI.

1.

2.

3.

LNG‐20 versus non‐hormonal ≤250mm2 IUDs

3. LNG‐20 IUS versus IUD <=250mm2: Rate ratios with 95% CI.

4. LNG‐20 IUS versus IUD<=250mm2: Life table differences with 95%CI.

4.

5.

LNG‐20 versus stainless steel ring IUD

LNG‐20 versus subdermal implants

5. LNG‐20 IUS versus subdermal implants: Rate ratios with 95% CI.

LNG‐20 versus combined oral contraceptive

6. LNG‐20 IUS versus oral contraceptives: Rate ratios with 95% CI.

P4‐IUS versus non‐hormonal IUDs <=250mm2

7. Progestasert versus IUD <=250mm2: Rate ratios with 95% CI.

8. Progestasert versus non‐medicated IUD: Rate ratios with 95% CI.

Discussion

Authors' conclusions

Implications for practice.

Implications for research.

What's new

History

Acknowledgements

Data and analyses

Comparison 1. LNG‐20 IUS vs. IUDs >250mm2.

1.1. Analysis.

1.2. Analysis.

1.3. Analysis.

1.4. Analysis.

1.5. Analysis.

1.6. Analysis.

1.7. Analysis.

1.8. Analysis.

1.9. Analysis.

1.10. Analysis.

1.11. Analysis.

1.12. Analysis.

1.13. Analysis.

1.14. Analysis.

1.15. Analysis.

Methods	Setting: Chile, Maternity Unit   400 women randomised   Follow up: 1 year
Participants	Postpartum
Interventions	P4‐IUS [n=200] vs. CuT 200 IUD [n=200] ‐ 100 inserted by hand and 100 inserted an inserter in each group
Outcomes	Pregnancy   Continuation   Menstrual change
Notes	Quality assessment:   Randomisation technique: No mention   Allocation concealment technique: No mention   Description of prior contraceptive method / pregnancy provided   Measurement: Groups treated identically   Method of analysis: Other
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	Setting: Finland and Brazil, Family Planning Clinics   484 women randomised   Follow up: 2 years (Brazil and Finland) and 5 years (Finland only)
Participants	18‐40 years   Proven fertility   Not breast feeding
Interventions	LNG‐20 and LNG‐30 IUSs [n=164 and 163, respectively] vs. Nova‐T IUD [n=157]
Outcomes	Pregnancy   Continuation   Resaons for discontinuation   Hormonal side effects   Menstrual change
Notes	Quality assessment:   Randomisation technique: Random tables (permutations of nine numbers)   Allocation concealment technique: No mention   Description of prior contraceptive method / pregnancy provided   Double‐blinded assessment of outcomes   Measurement: Groups treated identically   Method of analysis: Pearl indices and life tables (multiple and single decrement rates)   User / method failure reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	Setting: Clinics (4)   676 women randomised   Follow up: 1 year
Participants	Various parity
Interventions	P4‐IUS [n=359] vs. inert IUD [n=317]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation   Menstrual change
Notes	Quality assessment:   Randomisation technique: No mention   Allocation concealment: 'both types of device were externally identical'   Double‐blinded assessment of outcomes   Measurement: Groups treated identically   Method of analysis: Life tables
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	Setting: Finland, Family Planning Clinics   298 women randomised   Follow up: 1 year
Participants	18‐43 years   Variable parity   Regular menses
Interventions	LNG‐20 IUS [n=147] vs. LNG‐20 ICD [n=151]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation   Hormonal side effects
Notes	Quality assessment:   Randomisation technique: Random number table with group allocation predetermined   Allocation concealment technique: Consecutively numbered opaque sealed envelopes opened just before IUS insertion   Measurement: Groups treated identically   Method of analysis: Life tables (single decrement rates)   User / method failure reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Methods	Setting: Demark, Finland, Hungary, Norway, Sweden, family planning clinic 438 women randomised 2:1 as a segment of a larger trial of 3000 Follow‐up: 1, 3 and 5 years
Participants	post elective termination
Interventions	Mirena [n=305] vs. Nova T [n=133]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation
Notes	Quality assessment:   Randomisation technique: not mentioned   Allocation concealment: Sealed envelopes   Measurement: Groups treated identically   Method of analysis: Other data
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Methods	Setting: Chile, Family Planning Clinics   295 women randomised   Follow up: 1 year
Participants	17‐40 years   Parous   Regular menses
Interventions	Progesterone T IUS [n=146] vs. Cu 7 IUD [n=149]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation   Menstrual change
Notes	Quality assessment:   Randomisation technique: Computed tables   Allocation concealment technique: No mention   Description of prior contraceptive method / pregnancy reported   Blinded assessment of outcomes   Measurement: Groups treated identically   Method of analysis: Life tables (method not stated)   User / method failure reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	Setting: Moscow Russia, Ob/Gyn outpatient department 60 women randomised Follow‐up: 1 year
Participants	18‐45, with controlled insulin‐dependent diabetes mellitus Parous
Interventions	Levonorgestrel intrauterine system [n=26] vs. copper T 380A [n=28]
Outcomes	Continuation
Notes	Quality assessment:   Randomisation technique: Computer generated random numbers   Allocation concealment technique: Sealed opaque envelopes opened in numerical order   Description of prior contraceptive method / pregnancy reported   Women blind to method   Measurement: Groups treated identically   Method of analysis: Other
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Methods	Setting: France   Follow up: < 1 year   30 women randomised
Participants	24‐42 years   Multiparous
Interventions	P4‐IUS [n=13] vs. CuT 200 IUD [n=17]
Outcomes	Pregnancy   Menstrual change and blood loss
Notes	Quality assessment:   Randomisation technique: No mention   Allocation concealment technique: No mention   Measurement: Groups treated identically   Method of analysis: Other
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	D ‐ Not used

Methods	Setting: Egypt, Department of Obstetrics and Gynecology 320 women randomised Follow‐up: 1 year
Participants	Parous Exclusively breastfeeding for at least 1 year
Interventions	LNG‐IUS [n=163] vs. Cu T380A IUD [n=157]
Outcomes	Continuation
Notes	Quality assessment:   Randomisation technique: Computer generated sequential numbers   Allocation concealment: Sealed opaque envelopes opened in sequential orderMeasurement: Groups treated identically   Method of analysis:Other
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Methods	Setting: Multinational (Singapore, Brazil, Egypt and USA), Family Planning Clinics   2226 women randomised   Follow up: 7 years
Participants	18‐38 years   Parous
Interventions	LNG‐20 IUS [n=1125] vs. CuT 380Ag IUD [n=1121]
Outcomes	Pregnancy   Continuation   Reasons for discontinuation   Insertion problems   Hormonal side effects   Menstrual change   Adverse events   Pregnancy after discontinuation of method
Notes	Quality assessment:   Randomisation technique: Random numbers ‐ blocks of 50   Allocation concealment: Sealed opaque envelopes opened in ascending numerical order   Women blinded to method   Measurement: Groups treated identically   Method of analysis: Life tables (multiple and single decrement rates)   User / method failure reported   Active follow up conducted
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate