Correspondence: Dr T. Backman, Department of Gynaecology and Obstetrics, University Hospital Turku, P.O.B. 52, FIN-20510 Turku, Finland.
Objective To establish the continuation rates of the levonorgestrel intrauterine system (LNG IUS) and symptoms associated with its premature removal.
Sample and Setting All women in Finland who had a LNG IUS inserted between April 1990 and December 1993 and whose doctor had filled in and returned a form at the insertion visit. This study population consists of 46% of all the LNG IUSs sold in Finland between 1990 and 1993.
Design A questionnaire on reproductive and contraceptive history, gynaecological problems and symptoms experienced during the use of the LNG IUS was sent to 23,885 LNG IUS users. A total of 17,914 questionnaires were returned (response rate 75%). The results cover experience from 58,600 woman years. A log-rank-test was used to test differences in continuation rates. Multivariate analyse were performed using Cox's proportional hazard model.
Results The LNG IUS was prematurely removed from 5175 women. The one, two, three, four and five year continuation rates were 93%, 87%, 81%, 75% and 65%, respectively. The symptoms during the use of the LNG IUS most strongly associated with its premature removal were excessive bleeding and spotting, and infections and pain. The risk of premature removal was markedly lower among women who had occasional or total absence of menstruation. Premature removal was less likely in the oldest age group.
Conclusions The continuation rate of the LNG IUS compares favourably with other long-acting contraceptive systems. Totally or occasionally absent menstruation was strongly associated with prolonged continuation.
The levonorgestrel intrauterine system (LNG IUS) (Mirena, Levonova Leiras, Finland) is a long acting intrauterine contraceptive system that releases levonorgestrel into the uterine cavity at a rate of 20 μ/d1. The effective life-time of the LNG IUS has been shown to be five years2,3. It combines the positive effects of both hormonal and intrauterine contraception. The contraceptive efficacy and safety of the LNG IUS have been presented in randomised comparative trials4–8.
The continuation rates for other reversible long-acting contraceptive methods, mainly copper intrauterine devices and levonorgestrel-releasing implants (Norplant, Leiras, Turku, Finland), have generally been good as demonstrated in large comparative trials9–12. The major medical reasons for discontinuation have been bleeding disorders and pain2,4–6,8.
The objective of this study was to determine the overall continuation rates of the LNG IUS and to evaluate the symptoms associated with the removal of the system. Even though randomised clinical studies have given ample evidence of the safety and efficacy of the LNG IUS, an epidemiological survey was designed to establish the continuation rates and symptoms related to discontinuation among a large group of LNG IUS users.
The LNG IUS received marketing authorisation in Finland in 1990. Concomitantly with the launch of the product in Finland, a post marketing survey was started in 1990. The women who had a LNG IUS inserted between April 1990 and December 1993 were asked to participate in the survey. A short questionnaire was filled in at insertion. The women consented to the registration of their social security codes by signing their initials. A total of 26,630 insertion forms were returned to the manufacturer. This represented 46% of all the LNG IUSs sold between April 1990 and December 1993 in Finland.
The names and addresses of the women were received from the Population Register Center using the social security codes gathered at the insertion visit. Approval for the study was obtained from the Joint Ethics Committee of Turku University and Turku University Hospital. The names and addresses of 2445 women were not obtained.
A questionnaire with 75 questions was mailed to 23,885 women in April 1996. The mailing system and the questionnaire were tested in two pilot studies on 300 women. After three reminders a total of 17,914 questionnaires were returned, which represented 75% of the mailed forms.
By signing the questionnaire, the women consented to the use of the information gained and to later collection of more information from national registers. The date of birth was also asked for. All unsigned questionnaires and the ones in which the date of birth did not match with the social security code used were excluded from the final analysis (n= 554). The final number of applicable questionnaires was 17,360 (Fig. 1).
The mean age of the women at the time of answering the questionnaire was 36 years. 53% had two children, and the number of nulliparous women was only 112 (Table 1).
Table 1. Age groups and parity. Values are given as n (%).
*Frequency missing 327.
The women were asked about their general health status, reproductive and contraceptive history and gynaecological problems. They were asked whether they had experienced the following symptoms or events at any time during the use of the LNG IUS: excessive bleeding, irregular menstruation, occasional or total absence of menstruation, pregnancy, recurrent vaginal infections, abdominal pain, headache, swelling of breasts or lower abdomen, weight gain, tenderness of breasts, excessive hairiness, loss of hair, greasiness of skin or hair, pelvic inflammatory disease, spotting, depression and sexual disinterest. The women were asked to answer with either ‘yes’ or ‘no’. The frequency or duration of the symptoms was not asked, neither was the time period during the use of the LNG IUS at which the symptom was experienced.
Women who had had their LNG IUS removed because they were planning pregnancy were excluded from the present analysis (n= 881). Women over 48 years of age were also excluded (n= 248). Hence, the number of applicable questionnaires was 16,231 (Fig. 1).
Age, socioeconomic and educational status of the respondents and nonrespondents were compared with data obtained from the Central Statistical Office of Finland. The respondents were more educated and more often in employment than the nonrespondents (Table 2).
Table 2. The baseline variables of the respondents and nonrespondents compared with the general population. Values are given as %.
In working life
Basic level education
The continuation time was the time between the insertion date and the date of removal or the date of signing the questionnaire. The date of insertion was collected from the form filled in at the insertion visit, and the date of removal was asked for in the questionnaire. Either the date of signing the questionnaire or the date of removal of the LNG IUS was used as the cutoff date. The analyses were interrupted at five years. This information was used to calculate the continuation rate for the LNG IUS.
A log-rank-test was used to test differences in continuation between subgroups of women. A multivariate analysis was performed using Cox's proportional hazard model to predict the premature removal of the LNG IUS. Differences between groups were evaluated using risk ratios and 95% confidence intervals (CI). A risk ratio (RR) smaller than one indicates a prolonged use of LNG IUS.
The LNG IUS was removed from 5175 women by the time of signing the questionnaire. The continuation rate was highest in the oldest age group (Fig. 2). The one, two, three, four and five year continuation rates for the LNG IUS were 93%, 87%, 81%, 75% and 65%, respectively (Fig. 3).
A total of 108 women reported an unplanned pregnancy during the use of the LNG IUS. Of these, 44 were ectopic. The Pearl Index over five years of use is 0.18. The frequency of symptoms reported during the use of the LNG IUS is shown in Table 3 and the significant risk ratios for premature removal in Table 4.
Table 3. The symptoms reported during the use of the levonorgestrel intrauterine system included in the final analysis. Values are given as %
< 33 years
Totally or occasionally missed period
Swelling of breasts or the lower abdomen
Tenderness of breasts
Loss of hair
Pelvic inflammatory disease
Table 4. Risk ratios and 95% confidence intervals for premature removal.
Symptoms during the use of Mirena
Univariate analyse including age *†
*Only symptoms with statistical significance have been included in this table.
†The study population was divided into three age groups: < 33,33–38, 3943 years, the division was based on the equal size of the three groups.
‡Additional factors included in the multivariate analysis: age, marital status, body mass index, chronic diseases, parity, the use of alcohol, abortions, miscarriages, and severe gynaecological infections or ovarian cysts during the use of the system.
Pelvic inflammatory disease
Totally or occasionally missed period
Loss of hair (scalp)
Excessive hairiness (body)
Greasiness of hair and skin
Tenderness of breasts
A risk ratio for premature removal of the LNG IUS was first estimated separately for each symptom experienced during the use of the LNG IUS.
Weight gain, swelling of the breasts or lower abdomen, headache and sexual disinterest had no statistically significant association with premature removal of the LNG IUS, and are therefore not shown in Table 4.
In the multivariate model, bleeding disorders, consisting of spotting and excessive bleeding, had the strongest association with the premature removal of the LNG IUS (RR 2.77, 95% CI 2.51–3.07). The following symptoms, in descending order, were also associated with a significantly higher risk of premature removal: pelvic infections (RR 1.40, 1.25–1.57), pain (RR 1.32, 1.23–1.42), depression (RR 1.33, 1.24–1.43) and recurrent vaginal infections (RR 1.25, 1.14–1.38) (Table 4).
The risk of premature removal of the LNG IUS was significantly lower in women who had an occasional or total absence of menstruation (RR 0.46, 0.43–0.50). Hormonal adverse effects, such as loss of hair of the scalp, excessive hairiness of the body, greasiness of hair and skin and tenderness of the breasts, were associated with prolonged use of the LNG IUS.
The main objective of this study was to find out the continuation rate for the LNG IUS in normal use and the symptoms associated with its premature removal. Removals due to age > 48 years or planning pregnancy were not considered premature. Women planning pregnancy were excluded from the analysis. Symptoms experienced by women over the age of 48 may be menopausal and this group was also excluded from the analysis. This study involved 17,360 women, compared with 1821 women in the largest prospective clinical trial4. The large population makes the study unique among studies on contraception.
Nonresponders were less educated and less often in employment than responders (Table 2). According to population surveys people with lower socioeconomical status are less prone to respond11, so this is unlikely to bias the results.
The proportion of nulliparous women was very low (0.7%). This was expected since the manufacturer recommends the LNG IUS primarily for parous women. The Pearl Index over five years use is 0.18, which compares well with the figures in earlier studies4,13.
Long term continuation rates for copper IUDs and the LNG IUS in clinical trials have been 33–53%2,4,10,13, which are similar to the figures for the levonorgestrel releasing implant (Norplant)9,11,12.
The continuation rate in this study was as good as or even better in regular use than in earlier clinical trials2,4,13. Even if women planning pregnancy were included in the analysis, the continuation rates compare well with those in clinical trials (Fig. 3).
In clinical trials, the cumulative discontinuation rate due to amenorrhoea has been reported to be from 1.5 to 4%13. In the oldest studies, amenorrhoea (no bleeding or spotting during three months) has accounted for still higher risk of removal2,12,14. When women were adequately counselled about the benign nature of totally or occasionally missed menstruation, they considered the absence of menstruation an advantage and thus in the most recent clinical studies removal rates due to amenorrhea have been low4. In fact, the women with regular menstruation have been observed to discontinue the use of a contraceptive method more often than those with delayed bleeding episodes or oligomenorrhea15.
If excessive bleeding, spotting or pelvic inflammatory disease was experienced during the use of the LNG IUS, there was a clearly higher risk of having it removed prematurely. In the earlier studies, the most common reasons for discontinuing the use of intrauterine or other continuous progestin-only contraceptives have also been bleeding disorders and pain9. This finding was confirmed in this study.
The frequency of symptoms characterised as hormonal was much higher in this survey than in earlier clinical trials. However, the removal rate due to hormonal symptoms is similar to reported results in earlier studies2,4,7,13. The symptoms asked about are very common and can be assumed to be experienced by many women regardless of the method of contraception.
In earlier studies the gross cumulative removal rate due to hormonal adverse effects such as loss of hair, excessive bodily hair, greasiness of hair and skin or tenderness of the breasts alone has been low, but higher than with copper IUDs4,13. As in earlier studies, the hormonal adverse effects seldom lead to premature removal of the LNG13,16.
In agreement with this finding, the results of this study indicate that these adverse effects are related with continued use of the LNG IUS. This may suggest that these symptoms appear with time. Randomised clinical trials have, however, shown a higher occurrence of these symptoms during the first year compared with the subsequent years of use4. The finding may partly be explained by a suggestion that these symptoms have low impact on women's lives and therefore do not lead to alterations on contraceptive methods.
The present study of 17,360 LNG IUS users confirms the finding of clinical trials that the continuation rate for the LNG IUS is high for up to five years. Occasionally or totally missed menstruation is associated with prolonged/continued use. Symptoms associated with premature removal of the system are bleeding disorders, pain and pelvic infections.
We wish to thank Mr J. Tuominen for his valuable assistance in the statistical analysis. This study has been financially supported by Leiras Oy, Turku, Finland.