Levonorgestrel-releasing (20 μg/day) intrauterine systems (Mirena) compared with other methods of reversible contraceptives

Authors

  • R. S. French,

    Research Fellow, Corresponding author
    1. Department of Sexually Transmitted Diseases, Royal Free and University College London Medical School
      Correspondence: Dr R. French, Department of Sexually Transmitted Diseases, Royal Free and University College Medical School, The Mortimer Market Centre, off Capper Street, London WC1E 6AU, UK.
    Search for more papers by this author
  • F. M. Cowan,

    Senior Lecturer (Genitourinary Medicine)
    1. Department of Sexually Transmitted Diseases, Royal Free and University College London Medical School
    Search for more papers by this author
  • D. Mansour,

    Consultant (Community Gynaecology & Reproductive Health Care)
    1. Graingerville Contraception and Sexual Health Clinic, Newcastle City Health NHS Trust
    Search for more papers by this author
  • J. P. T. Higgins,

    Research Fellow (Medical Statistics)
    1. Systematic Reviews Training Unit, Institute of Child Health, Royal Free and University College London Medical School
    Search for more papers by this author
  • A. Robinson,

    Consultant (Genitourinary Medicine)
    1. Department of Genitourinary Medicine, Camden & Islington Community Health Services NHS Trust
    Search for more papers by this author
  • T. Procter,

    (Deputy Chief Officer)
    1. Islington Community Health Care Council
    Search for more papers by this author
  • S. Morris,

    Lecturer (Economics)
    1. Department of Health Economics, School of Social & Human Sciences, City University, London
    Search for more papers by this author
  • J. Guillebaud

    Professor
    1. Margaret Pyke Family Planning Centre, Camden & Islington Health Services NHS Trust, London
    Search for more papers by this author

Correspondence: Dr R. French, Department of Sexually Transmitted Diseases, Royal Free and University College Medical School, The Mortimer Market Centre, off Capper Street, London WC1E 6AU, UK.

Abstract

Objective To assess the relative contraceptive effectiveness, tolerability and acceptability of the levonorgestrel-releasing (20 μg per day) intrauterine system (LNG-20) compared with reversible contraceptive methods in women of reproductive age.

Design A systematic review and meta-analysis of randomised controlled trials.

Identification Studies were identified through seven databases, and by contacting investigators and organisations working in the contraceptive field.

Main outcome measures Unplanned pregnancy and continuation of contraceptive method.

Results Five of the seven randomised controlled trials which met the inclusion criteria were included in the meta-analyses; four were comparisons of the LNG-20 intrauterine system with nonhormonal intrauterine devices. LNG-20 intrauterine systems were compared with intrauterine devices divided into two categories, those > 250 mm3 (Copper T 380 Ag and Copper T 380 A intrauterine devices) and those ≤ 250 mm3 (Nova-T, Copper T 220C and Copper 200 intrauterine devices). Pregnancy rates for the LNG-20 intrauterine system users were significantly less likely to become pregnant compared with users of intrauterine devices ≤ 250 mm3, and significantly less likely to have an ectopic pregnancy. LNG-20 intrauterine system users were more likely to experience amenorrhoea and device expulsion than women using intrauterine devices > 250 mm3. LNG-20 intrauterine system users were significantly more likely than all the intrauterine device users to discontinue because of hormonal side effects and amenorrhoea. When the LNG-20 intrauterine system was compared with Norplant-2, the LNG-20 users were significantly more likely to experience oligo-amenorrhoea, but significantly less likely to experience prolonged bleeding and spotting.

Conclusions The effectiveness of the LNG-20 intrauterine system was similar to or better than other contraceptive methods with which it was compared. Amenorrhoea was the main reason for the discontinuation of the LNG-20 intrauterine system, usually unnecessarily, since this end-organ suppression of bleeding is benign, associated with normal oestrogen levels. Women choosing this method should be informed of potential amenorrhoea when having pre-contraceptive counselling and that absent bleeding may be viewed as a positive outcome.

INTRODUCTION

Since 1970 research into several hormonally medicated intrauterine systems led to the development of the levonorgestrel-releasing intrauterine system (levonorgestrel-20 intrauterine system or Mirena). Licensed in 25 countries for five years of contraception1, the levonorgestrel-20 intrauterine system has a 32 mm lóng T-shaped plastic frame with a reservoir on the vertical stem of the intrauterine system containing 52 mg of levonorgestrel mixed with polydimethylsiloxane2. This allows a steady, local release of 20 μg levonorgestrel per day through the rate-limiting surface membrane.

The levonorgestrel-20 intrauterine system is more expensive than copper-bearing intrauterine devices (net cost £99 compared with approximately £9). It offers, however, considerable gynaecological benefits, particularly in women with heavy or painful menses. The levonorgestrel-20 intrauterine system has been demonstrated to reduce menstrual loss and is a rational alternative to hysterectomy for many women with heavy menstrual bleeding3,4.

The objectives of this systematic review were to assess the contraceptive efficacy, tolerability and acceptability of the levonorgestrel-20 intrauterine system, compared with other reversible contraceptive methods, in women of reproductive age. The work presented here is part of a wider systematic review looking at all hormonally impregnated implantable contraceptives5.

METHODS

All randomised controlled trials which met the following criteria were included:

  • 1Study participants were women of reproductive years.
  • 2Comparisons of the levonorgestrel-20 intrauterine system with another reversible contraceptive (intrauterine devices, barrier contraceptives, oral contraceptives, injectable contraceptives and other subdermal implants).
  • 1Predetermined outcomes reported. The primary outcome measures were pregnancy due to method or user failure and continuation of contraceptive method.

All publications of studies comparing levonorgestrelintrauterine systems with other reversible contraceptive methods were sought through computerised searches in the Cochrane Controlled Trials Register, MEDLINE, POPLINE, EMBASE, PsycLIT and specialist databases of family planning libraries from January 1974 to July 1998. All reference lists of located papers were checked to identify further studies. Requests for unpublished data were made to individuals and organisations in the contraceptive field.

All papers were blinded for author and journal. Data were extracted independently by two reviewers (R.F. and F.C.) onto data collection forms. Single-decrement life table probabilities with their standard errors (SE) and events per women months (akin to the Pearl Index rate6) were collected for each outcome at specific follow up points (i.e. one, two, three, four and five years). Definitions of contraceptive statistical measures are provided in Table 1. All life table analyses except continuation probabilities in this paper are based on single decrement life tables unless otherwise stated. Differences were resolved by discussion and consensus. Data were extracted from non-English articles by one of the reviewers (R.F.) with a translator. Quality was assessed independently in terms of general methodological factors which may bias study results, as well as some of contraception specific factors recommended by Trussell et al.7

Table 1.  Definition of contraceptive measures and methods of data synthesis. PI = Pearl index.
MeasureDefinition
  1. *Rate ratios were calculated rather than risk ratios as in contraceptive literature the denominator is usually women months or years instead of the number of women.

Pearl indexPI = No. of events/total no. of women months .1 200 (1 300 if menstrual cycle used). Its disadvantage is that variations of risk over time are not accounted for (e.g. those at high risk of pregnancy have a greater chance of pregnancy in the early stages of follow up) and variations in loss to follow up are not accounted for. Measure for meta-analysis: The ratio of rates* (i.e. number of events divided by total number of women months) for the two groups of users were calculated7 to give a relative measure of treatment effect.
Multiple decrement life tablesMeasure of survival (or failure) of individuals or groups over time. Frequently used in contraceptive research to measure the probability of pregnancy or method continuation. These measure multiple reasons for discontinuation, including pregnancy, hormonal side effects, etc. Also known as net, crude or competing rates.
Single decrement life tablesCalculated the same way as multiple decrement life tables, but at the time of discontinuation they censor individuals who stop using a method for reasons other than the one being measured and are therefore recommended as the most appropriate method to measure contraceptive outcomes. Also known as gross, net or noncompeting rates. Measure for meta-analysis: The difference between single-decrement life table probabilities for events for the two groups of users was calculated. This gives an absolute difference in percentage terms of treatment effect (i.e. attributable risk).

Few studies reported contraceptive effectiveness using the gold standard of single decrement life tables10 Therefore two measures of effect size were collected (Table 1). The exceptions were for planned pregnancy after discontinuation of contraceptive method, hormonal side effects and menstrual outcomes, where relative risks were calculated8. Menstrual changes were reported at three monthly intervals as recommended in the contraceptive literature9.

Data from different studies were only combined if collected over the same time periods and when the comparative interventions were similar, such as the levonorgestrel-20 intrauterine system versus nonhormonal intrauterine devices or versus subdermal implants. Copper-medicated intrauterine devices were divided into two categories, based on the surface area of the copper wire. The first, defined as intrauterine devices > 250 mm3, included CuT 380 A and CuT 380 Ag, and the second, defined as ≤ 250 mm3, included the Nova-T, CuT 200 and CuT 220 intrauterine devices.

A random effects approach was used for the metaanalysis to account for variability between studies11. In the absence of heterogeneity this coincides with a fixed effect analysis. No heterogeneity was identified in the analyses unless explicitly stated.

RESULTS

The search strategy identified over 400 publications. Seven randomised controlled trials met the inclusion criteria, reported in 23 publications11–33.

Three studies20,34,35 compared the levonorgestrel-20 intrauterine system with intrauterine devices > 250 mm3 and three15,27,34 compared the levonorgestrel-20 intrauterine system with intrauterine devices > 250 mm3. The two remaining comparisons were the levonorgestrel-20 intrauterine system versus Norplant rods31 and the levonorgestrel-20 intrauterine system versus the levonorgestrel-intracervical device12. Five studies reported that all participants had proven fertility12,20,27,31,34. The study by Sivin et al.20 was the longest, with seven years follow up.

Information was also collected on additional factors, which could potentially affect the results. Two of the seven randomised controlled trials reported that women had received contraceptive counselling14,31. Five studies provided information on timing of insertion during the menstrual cycle14,20,27,31,34, all within appropriate time frames. In one study20 physicians found insertion of the levonorgestrel-20 intrauterine system more difficult than insertion of the CuT 380 Ag intrauterine device and more women reported moderate to severe pain at the insertion of the levonorgestrel-20 intrauterine system.

The quality factors with most discrepancies between the reviewers were whether allocation concealment had been adequately described, and the method of statistical analysis for pregnancy and continuation outcomes (the results of the quality assessment are available from the authors).

It was possible to extract appropriate data from five of the seven identified randomised controlled trials14,20,27,31,34. In the remaining two12,35 neither the number of woman months of follow up nor the standard errors of life table probabilities were reported.

There was no evidence to suggest that the levonorgestrel-20 intrauterine system was relatively more or less effective in preventing unwanted pregnancy than intrauterine devices > 250 mm3 (Fig. 1). No significant difference in contraceptive effect was found in the pooled life table probability difference20,34.

Figure 1.

Pregnancy: levonorgestrel-releasing (20μg per day) intrauterine system (LNG-20 IUS) versus intrauterine devices (IUD) > 250 mm3.

However, the levonorgestrel-20 intrauterine system was significantly more effective in preventing pregnancy than intrauterine devices ≤ 250 mm3 at all follow up points (Fig. 2). There was evidence of heterogeneity between the two studies at year one, while the fixed and random effects analyses did not coincide. The direction effect, however, was the same so we are not unduly concerned by this.

Figure 2.

Pregnancy: levonorgestrel-releasing (20 μg per day) intrauterine system (LNG-20 IUS) versus intrauterine devices (IUD) ≤ 250 mm3.

The summary rate ratios suggest continuation of the levonorgestrel-20 intrauterine system were similar to the other nonhormonal intrauterine devices (Figs. 3 and 4). However, the life table differences calculated from one study20 suggests that levonorgestrel-20 intrauterine system users are significantly less likely to continue when compared with users of the CuT 380 Ag intrauterine device. The absolute reductions in continuation probabilities over one, two and five years were 6.3% (10% to 2.6%), 8.1% (12.4% to 3.8%) and 7.6% (11.9% to 3.3%), respectively.

Figure 3.

Continuation: levonorgestrel-releasing (20 μg per day) intrauterine system (LNG-20 IUS) versus intrauterine devices (IUD) > 250 mm3.

Figure 4.

Continuation: levonorgestrel-releasing (20 μg per day) intrauterine system (LNG-20 IUS) versus intrauterine devices (IUD) ≤ 250 mm3.

There were no significant differences in the relative risks for pregnancy after removal of the levonorgestrel-20 intrauterine system compared with nonhormonal intrauterine devices16,26.

It was only possible to extract data on some hormonal side effect outcomes from the Andersson et al. study14, which compared the levonorgestrel-intrauterine system with the Nova-T intrauterine device. No significant differences were observed. At five years the relative risk (95% CI) for ovarian cysts was 1.5 (0.51 to 4.40), for headaches 1.71 (0.49 to 6.02), for breast tenderness 1.50 (0.31 to 7.17), for acne 5.56 (0.73 to 42.35), and for nausea 5.0 (0.24 to 103.86). However, Luukkainen et al.21 observed that users of the levonorgestrel-20 intrauterine system were significantly more likely to report an increase in headaches and acne compared to the Nova-T intrauterine device users.

One study20 indicated that levonorgestrel-20 intrauterine system users were significantly more likely to experience amenorrhoea compared to women using the CuT 380 Ag intrauterine device. This increased over time. At three months the relative risk (95% CI) was 2.25 (1.3 to 3.56), increasing to 7.24 (4.14 to 12.55) at year three. No significant differences were noticed in prolonged bleeding. It was not possible to extract data for any other of the specified menstrual outcomes, but the study by Sivin et al.20 also reported that levonorgestrel-20 intrauterine system users were significantly less likely to experience dysmenorrhoea.

Levonorgestrel-20 users were no more or less likely to experience expulsion of device than intrauterine device > 250 mm3 users after one year. After five years over 50% of users were more likely to have expelled the device, [rate ratio 1.53 (1.13 to 2.07)]20. This was also reflected in summary difference calculated between life tables20,34.

The rate ratios indicate that women using the levonorgestrel-20 intrauterine system are significantly less likely to have had an expulsion at two years of use than intrauterine device ≤ 250 mm3 users. The rate ratios (95% CI) at one, two and five years of follow up are 0.97 (0.63 to 1.48)14, 0.11 (0.02 to 0.60)27 and 0.27 (0.06 to 1.13)27, respectively. Life table probability differences from another study34 were not significant. There was no significant difference in the rate of devices becoming embedded20, [rate ratio (95% CI) of 7.0 (0.36 to 135.52)].

One study comparing the levonorgestrel-20 intrauterine system with the CuT 380 Ag intrauterine device20 and two comparing the levonorgestrel-intrauterine system with the Nova-T intrauterine device14,27 provided sufficient information to calculate rate ratios for ectopic pregnancy. No significant differences were found for the levonorgestrel-20 intrauterine device versus CuT 380 Ag comparison. After one year there was no significant difference for the levonorgestrel-20 intrauterine system versus Nova-T intrauterine device comparison14,27, but after three years the rate ratio was 0.1 (0.02 to 0.62) and after five years 0.07 (0.01 to 0.41). No significant differences were observed in the rates of pelvic inflammatory disease20,27. The study by Andersson et al.14 observed that levonorgestrel-20 intrauterine system users were significantly less likely to experience pelvic inflammatory disease, particularly younger women, but we were unable to use the data in the meta-analysis.

In general, levonorgestrel-20 intrauterine system users were significantly more likely to discontinue through hormonal side effects than nonhormonal intrauterine device users. When compared with intrauterine devices > 250 mm3 a significant difference was not seen until five years. The rate ratios (95% CI) at one, three and five years were 0.81 (0.23 to 2.80)20, 1.71 (0.64 to 4.55)34, and 4.24 (1.99 to 9.04)20, respectively. One study20, giving life table probabilities, found no significant difference at one year. When compared with intrauterine devices ≤ 250 mm3, the rate ratios after one, three and five years of use were 5.40 (2.25 to 12.97)14, 3.05 (0.24 to 38.34)14,34 and 5.18 (1.32 to 20.34)27, respectively. Heterogeneity was identified between the two studies in the random effects analysis at year three. This was explained by the opposing directions of the rate ratios: 11.2 (4.77–26.42) in the study by Andersson et al.14 and 0.85 (0.42–1.73) in the study by Baveja et al.34.

Levonorgestrel-20 intrauterine system users were significantly more likely to discontinue for menstrual disturbance than women using intrauterine devices > 250 mm3. Rate ratios (95% CI) were 1.48 (1.02 to 2.14) after one year and 1.48 (1.23 to 1.79) after five years20. In the life table differences there were absolute increases of 6.9% (2.9% to 10.9%)20,34, 11.1% (6.3 to 15.9%)34 and 14.5% (8.8% to 20.2%)34 after one, two and three years, respectively. Further investigation showed that amenorrhoea was the main cause for the levonorgestrel-20 intrauterine system discontinuations. The discontinuation rate ratios for amenorrhoea were 65.1 (4.01 to 109.84) and 48.92 (16.93 to 141.36) after one and five years20, respectively. Again, in the life table difference there was an absolute increase in the probability of discontinuation through amenorrhoea of 5.0% (3.2% to 6.9%) after one year20,34 and rising to 13.5% (9.3% to 17.3%) after three years34.

In contrast, levonorgestrel-20 intrauterine system users were significantly less likely to discontinue because of menstrual bleeding and pain, with a rate ratio (95% CI) of 0.71 (0.56 to 0.89) and life table difference of −7.6% (−10.9% to −4.9%) after five years20.

In comparison with intrauterine device ≤ 250 mm3, levonorgestrel-20 intrauterine system users showed no significant difference in discontinuation for menstrual change27. However, discontinuations due to amenorrhoea were again significantly increased after five years: rate ratio 29.2 (1.75 to 488.04). Significant differences in overall menstrual disturbance explaining discontinuation of the levonorgestrel-20 intrauterine system were found between the life table probabilities, although the data came from a different study34. Focusing on amenorrhoea alone, there was an absolute increase of 5.1% (3.4% to 6.8%) at one year in levonorgestrel-20 users, rising to 9.8% (7.5% to 12.1%) and 13.6% (10.8% to 16.4%) at years two and three, respectively.

Unpublished data on discontinuations of the levonorgestrel-20 intrauterine system compared with the Nova-T for amenorrhoea in the European randomised controlled trial14 demonstrated a huge variation between the participating centres, ranging from a multiple decrement probability of 2.7 in Finland to 19.6 in Hungary (I. Rauramo, personal communication). No significant differences were found on other adverse outcomes, pregnancy planning and personal choice as reasons for discontinuation.

With regards to other reversible contraceptive methods the only randomised controlled trial which met the inclusion criteria was a comparison of the levonorgestrel-20 intrauterine system with Norplant-231. No significant differences were observed for the following outcomes: pregnancy, continuation, expulsion, ovarian cysts, and adverse events, menstrual disturbance, and local device problems as reasons for discontinuation. However, there were significant differences in the relative risk of reported menstrual disturbance. levonorgestrel-20 intrauterine system users were significantly more likely to experience oligo-amenorrhoea compared with Norplant-2 users, but less likely to experience spotting and prolonged bleeding.

DISCUSSION

There is no evidence to suggest that the levonorgestrel-20 intrauterine system is any more or less effective than intrauterine devices > 250 mm3. The available data indicated that levonorgestrel-20 intrauterine system is significantly more effective than intrauterine devices ≤ 250 mm3 in preventing intrauterine and extrauterine pregnancies. For both these comparisons there were few data available for the meta-analyses. Further comparative data on the effectiveness will become available from ongoing studies such as the World Health Organisation levonorgestrel-20 intrauterine system versus CuT 380 A intrauterine device multicentre trial12. It will be interesting to see the effect of these studies on the summary effect sizes. Very large numbers of women need to be recruited into trials of contraceptive effectiveness, where two highly effective interventions are being compared, to ensure that there is adequate power to detect significant differences.

There were no studies comparing intrauterine systems with contraceptive methods more reliant on user compliance, such as oral contraceptives, where increased relative effectiveness might be expected. This may reflect the characteristics of women who choose or are offered the intrauterine system or intrauterine device and/or those approached for inclusion in the randomised controlled trials.

Continuation rates for the levonorgestrel-20 intrauterine system were significantly lower than those for intrauterine devices > 250 mm3. If a contraceptive is discontinued prematurely its value as a method of contraception is greatly reduced, even if it is highly effective. This is particularly important if one method is relatively more expensive than another, as is the case of the levonorgestrel-20 intrauterine system when compared with nonhormonal intrauterine devices. In addition, women who discontinue with a contraceptive method may become unintentionally pregnant before starting another method.

Levonorgestrel-20 intrauterine system users were significantly more likely to discontinue because of amenorrhoea than nonhormonal intrauterine device users. Although intrauterine devices are often said to cause increased menstrual bleeding and pain, and are therefore thought to be unacceptable to users, this was less frequently a cause for discontinuation than discontinuation of the intrauterine system-associated amenorrhoea. Discontinuation due to menstrual disturbance per se is not an informative outcome, as the levonorgestrel-20 and intrauterine device < 250 mm3 comparison illustrates. Data on discontinuation due to amenorrhoea, bleeding and pain must be collected separately to provide the true picture.

If women were counselled about potential for menstrual changes before the implantation, continuation might be higher. To date, there is little published evidence on the effectiveness of contraceptive counselling. Women who receive adequate precontraceptive counselling about the possibility of amenorrhoea or bleeding and pain may be dissuaded from these methods in the first place if these side effects are likely to concern them. More importantly, those who choose the levonorgestrel-20 intrauterine system maybe more likely to continue if they are made aware that they may experience amenorrhoea that has no adverse effects on their health. Certainly, unpublished European data from the Andersson et al. trial14 suggest discontinuation because of amenorrhoea is greatly influenced by nonbiological factors, such as differing cultures and provider attitudes in different countries (I. Rauramo, personal communication). The amenorrhoea in users of levonorgestrel-20 is benign and is due to high local concentrations of levonorgestrel in the endometrium. Research demonstrates that users have adequate ovarian follicular activity and the same mean oestradiol levels as copper intrauterine device users35.

In conclusion, this systematic review was unable to find any evidence that the levonorgestrel-20 intrauterine system was any more or less effective in preventing unwanted pregnancy than intrauterine devices > 250 mm3 and Norplant-2, but it is significantly more effective in preventing unwanted intrauterine and extrauterine pregnancies when compared with intrauterine devices ≤ 250 mm3. Women using the intrauterine system were more likely to experience amenorrhoea and this was a notable reason for discontinuation. Copper intrauterine devices > 250 mm3 are as effective as the levonorgestrel-20 intrauterine system and much cheaper. The CuT380, specifically, is licensed for 10 rather than five years. The levonorgestrel-20 intrauterine system should be offered to women who are concerned about menstrual bleeding and pain associated with intrauterine device use, or are experiencing heavy and painful menses. More research is required to assess the impact of counselling in relation to menstrual change on user satisfaction and on continuation rates.

Acknowledgement

Funding was provided by the NHS R&D Health Technology Assessment Programme.

Ancillary