Prospective study of the forearm bone mineral density of long-term users of the levonorgestrel-releasing intrauterine system

Abstract

BACKGROUND

The levonorgestrel-releasing intrauterine system (LNG-IUS) induces amenorrhoea, and its effect on bone mineral density (BMD) may constitute a concern. This study evaluated BMD in long-term users of the LNG-IUS or intrauterine device (IUD).

METHODS

BMD was evaluated at the midshaft of the ulna and ultra-distal radius using dual-energy X-ray absorptiometry in 37 women at 7 or 10 years of use following placement of a second LNG-IUS. The groups were paired for duration of use, age, body mass index (BMI), ethnicity and number of pregnancies.

RESULTS

The mean age of both LNG-IUS and IUD users at the 7th and 10th year was ∼34 and 38 years, respectively. Mean BMI was ∼25 in both groups, increasing to ∼26 at the 10th year. Amenorrhoea occurred in 51.4 and 91.9% of LNG-IUS users at the 7th and 10th year, respectively. Estradiol levels in LNG-IUS users were normal at both evaluations. There were no differences in BMD (g/cm²) at the midshaft of the ulna nor ultra-distal radius between LNG-IUS and IUD users or between the 7th and 10th years of use in LNG-IUS users. A Z-score below −2SD at the ultra-distal radius was observed in only one LNG-IUS user and in none of the IUD users at the 10th year. Higher BMI and BMD at the seventh year and amenorrhoea were predictors of higher BMD at the 10th year.

CONCLUSIONS

BMD at the midshaft of the ulna and ultra-distal radius in LNG-IUS users were similar to that of IUD users and remained unchanged between the 7th and the 10th years of use.

bone mineral density, levonorgestrel-releasing intrauterine system, Mirena, contraception

Introduction

Progestogen-only contraceptive methods are an attractive option for many women who require contraception for many years. However, the effect of current and past use on bone mineral density (BMD) is a medical concern among users of progestogen-only contraceptive methods (d'Arcangues, 2006), particularly in users of the injectable contraceptive depot-medroxyprogesterone acetate (DMPA; Guilbert et al., 2009) because of the transient hypoestrogenism that DMPA may induce (Bahamondes et al., 2000), this constituting the most important factor associated with bone loss (Bagger et al., 2004).

The levonorgestrel-releasing intrauterine system (LNG-IUS) is a highly effective contraceptive method with few side-effects and has been approved for use for up to 5 years (Luukkainen and Toivonen, 1995; World Health Organization, 2004). One of the characteristics of the method is that it induces amenorrhoea in a high percentage of women during use, this being provoked by the antiproliferative effect of LNG on the endometrium (Luukkainen and Toivonen, 1995; Ronnerdag and Odlind, 1999; Hidalgo et al., 2002). In view of this effect, the LNG-IUS has also been approved as a therapeutic option for the treatment of menorrhagia (Monteiro et al., 2002; Hurskainen et al., 2004; Kaunitzl et al., 2009) and as endometrial protection in post-menopausal women during estrogen therapy (Sitruk-Ware, 2007).

However, although much information is available on the contraceptive efficacy and health benefits of the LNG-IUS, up to the present time only one study has evaluated BMD in users of this method, a study that was carried out by our group (Bahamondes et al., 2006a). In that study, 53 non-obese women of 25–51 years of age (mean 34 years), who had been using the LNG-IUS for 7 years were compared with TCu380A intrauterine device (IUD) users paired for age (±1 year) and body mass index (BMI; kg/m²; ±1). The main findings were that estradiol (E₂) values were similar to those found in the follicular phase of the menstrual cycle and there were no significant differences in mean BMD at the midshaft of the ulna or at the ultra-distal radius between LNG-IUS and IUD users.

As stated above, amenorrhoea is common in users of the LNG-IUS; however, despite this characteristic, E₂ levels in users of the LNG-IUS are similar to those found during the follicular phase of the menstrual cycle (Luukkainen and Toivonen, 1995; Bahamondes et al., 2006a), even in long-term users of the same device (Hidalgo et al., 2009). Consequently, if serum E₂ levels are normal, the LNG-IUS would not be expected to affect BMD; however, this finding should be confirmed, particularly in long-term users of the device. Therefore, the aim of this study was to evaluate forearm BMD in users of a second consecutive LNG-IUS in whom BMD had been measured previously at the seventh year of use, and to compare findings with the BMD measurements of the same women at the 10th year of use. E₂ levels were also evaluated in these women at the same time-points. In addition, this cohort of women was compared with a group of women who had been using a TCu380A IUD for at least the same length of time and who served as a control group.

Materials and Methods

This prospective cohort study was conducted at the Human Reproduction Unit, Department of Obstetrics and Gynaecology, School of Medical Sciences, University of Campinas (UNICAMP), Campinas, Brazil. The Institutional Review Board approved the study, and all participants signed an informed consent form prior to admission.

The subjects consisted of a group of women in whom forearm BMD had been evaluated at the seventh year of use of an LNG-IUS (Bahamondes et al., 2006a), which was initially placed between April and September 1998 when these women agreed to participate in an acceptability study of the LNG-IUS (Mirena®, Bayer Schering Pharma Oy, Turku, Finland) in Brazil (Diaz et al., 2000; Hidalgo et al., 2002). At the end of the approved 5-year lifespan of the device, the women were invited to continue using the same device for a further 2 years. At the 7-year follow-up visit, the first 51 women who returned for a control evaluation underwent evaluation of forearm BMD as previously described (Bahamondes et al., 2006a). The women who had the LNG-IUS removed and a new one placed on the same day were invited to undergo another BMD evaluation in the 10th year of use. Therefore, 37 women of 25–48 years of age were evaluated in the 7th and 10th years of use. The non-users group consisted of 37 women who had been in use of the TCu380A IUD (Optima, Injeflex, São Paulo, Brazil) and who had never used any form of hormonal contraceptive. Women in the non-users group were paired with the LNG-IUS users according at least to their duration of use of the device, age (±1 year), BMI (kg/m²; ±1), ethnicity and number of pregnancies (±1). The IUD group were recruited for comparison at the seventh year of use and another group of IUD users were recruited for comparison at the 10th year of use.

Women were excluded from the study if they had previously used any hormonal contraceptive method or if they were currently taking calcium or vitamin D supplements, anticonvulsants, corticosteroids, thiazide diuretics or drugs for the treatment of thyroid disease. Women with chronic diseases such as diabetes mellitus, chronic renal failure, hyperthyroidism or hypothyroidism, hyperparathyroidism or hypoparathyroidism, hepatitis, cancer or pituitary diseases were also excluded. In addition, women were asked whether they had any parents or relatives who had osteoporosis or had suffered non-stress fractures and were excluded if they responded positively in order to avoid including women with hereditary factors for osteoporosis.

Definition of variables

BMD was the dependent variable and was defined as the relationship between bone mineral content (g/cm²) and the area of the bone measured. The independent variable was the use of the LNG-IUS or copper IUD at the 7th and 10th years. The control variables included age at the time of the first BMD measurement, ethnicity, number of pregnancies and deliveries, duration of exclusive and partial breastfeeding, weight, height, BMI (kg/m²), exercise habits, smoking habits and patterns of coffee and alcohol consumption.

BMD measurement

BMD was measured at the non-dominant forearm using dual-energy X-ray absorptiometry (DTX-200; Osteometer Meditech A/S, Rodovre, Denmark). Two BMD measurements were taken in each woman: (i) at the midshaft of the ulna (where the cortical bone predominates), at the point at which the radius is 8 mm from the ulna; and (ii) at the ultra-distal radius near the articulation with the bones of the carpus (where the trabecular bone predominates).

Estradiol measurement

Before performing the BMD evaluation, a blood sample was taken from a peripheral vein. After coagulation, the serum was separated and frozen at −20°C until being assayed. E₂ levels were measured by electrochemiluminescence immunoassay using a commercial kit (Roche Diagnostics GmbH, Mannheim, Germany) with a measuring range of 5.0–4300 pg/ml and inter-assay coefficient of variation of 4.9%. The normal values for follicular phase are 18.9–256.0 pg/ml. All samples were measured in duplicate.

Statistical analysis

Calculation of sample size was based on a previous study of BMD measurement carried out in users of the LNG-IUS (Bahamondes et al., 2006a) and a previous study of users of LNG- or etonogestrel-releasing contraceptive implants (Monteiro-Dantas et al., 2007), with an alpha of 5% and beta of 10%. Sample size was calculated at 37 women per group. Comparison between the groups of users and non-users with respect to demographic, anthropometric and obstetric variables and to BMD at both sections of the forearm was performed using Student's t-test for matched samples and the Wilcoxon signed rank test or Snedecor's test (ANOVA) as appropriate. Z-score values were compared using Fisher's exact test. In addition, a multiple regression analysis was performed to evaluate whether BMD was significantly associated with any variables between the 7th and 10th years of use. All data are presented as means ± standard error of the mean (SEM). The BMD values were presented as g/cm² and a Z-score (Consensus Development Conference, 1993).

Results

The 37 women in whom BMD was evaluated at the seventh year of use were re-evaluated at the 10th year of use. The mean duration of LNG-IUS use at the 7th and 10th years was 84.0 ± 0.0 and 123.0 ± 0.8 months (range 84–86 and 116–123 months), respectively, although women in the control group had been using the TCu380A IUD for a mean of 112.5 ± 4.2 months (range 84–204 months) and 131.2 ± 4.7 months (range 120–240 months) at the 7th and 10th years of evaluation, respectively. The sociodemographic and obstetric characteristics of the two study groups are shown in Table I. The mean age of the LNG-IUS and IUD users at the 7th and 10th years of use was ∼34 and 38 years, respectively (age range 25–47 and 28–51 years for the LNG-IUS users and 26–48 and 29–51 for the IUD users at the 7th and 10th years of use, respectively). BMI (kg/m²) at the same two time-points was ∼25, with a slight increase after 3 years of use to ∼26 in both groups of users. There were no significant differences between the two groups with respect to this variable or to any of the other characteristics. White women constituted ∼80% of participants in both groups at both times of evaluation, although smokers constituted ∼8 and 20% of users of the LNG-IUS and IUD groups, respectively. All smokers declared smoking 1–20 cigarettes per day (mean 8.3).

Table I

Selected characteristics of women in the two contraceptive groups at the 7th and 10th years of use (n = 37 pairs).

Variables	Duration of use
	7 years		P-value	10 years		P-value
	LNG-IUS¶	IUD@		LNG-IUS¶	IUD@
Age (years)	34.2 (0.94)	34.1 (0.93)	0.566*	37.9 (0.95)	37.9 (0.95)	0.689*
BMI (kg/m2)	24.8 (0.60)	24.9 (0.59)	0.603*	25.9 (0.59)	25.8 (0.57)	0.471*
Number of pregnancies	2.1 (0.17)	2.0 (0.16)	0.549•	2.1 (0.17)	2.4 (0.17)	0.033•
Duration of use (months)	84.0 (0.0)	112.5 (4.79)	<0.001•	123.0 (0.87)	131.2 (4.71)	0.322•
White women (%)	75.7	83.8	0.375#	75.7	75.7	>0.999#
Age at menarche (years)	12.9 (0.30)	12.8 (0.30)	0.952•	12.9 (0.30)	12.9 (0.24)	0.903•
Duration of breastfeeding (months)	9.2 (2.13)	8.2 (1.22)	0.880•	9.2 (2.13)	9.2 (1.62)	0.772•
Schooling not beyond elementary level (%)	62.2	54.1	0.664#	62.2	75.7	0.383#
Non-physical occupation (%)	62.2	67.6	0.815#	62.2	56.8	0.815#
Alcohol drinking (%)	2.7	16.2	0.125#	2.7	18.9	0.070#
Coffee drinking (%)	86.1	72.2	0.302#	86.5	94.6	0.375#
Tobacco use (%)	8.3	19.4	0.289#	8.1	24.3	0.109#

^¶LNG-IUS, levonorgestrel-releasing intrauterine system; ^@IUD, Intrauterine device; *Student's paired t-test; ^•Wilcoxon Signed Rank test; ^#McNemar paired test.

There were no significant differences between the two groups with respect to any of the other characteristics such as age at menarche, duration of exclusive breastfeeding, years of schooling, non-physical occupation or alcohol and coffee consumption (Table I). Amenorrhoea was reported by 19 (51.4%) and 34 (91.9%) of the 37 users of the LNG-IUS at the 7th and 10th years of evaluation, respectively (P < 0.001). In the other women, bleeding consisted of only a few episodes of spotting. Mean E₂ levels in LNG-IUS users, measured when BMD was evaluated, were 93.6 ± 7.7 (range 44.4–178.0) and 134.2 ± 28.7 (range 44.8–220.0) pg/ml at the 7th and 10th years of evaluation, respectively. All of the copper IUD users were menstruating regularly at both evaluation times.

Mean BMD (g/cm²) was 0.460 ± 0.009 at the midshaft of the ulna and 0.400 ± 0.011 at the ultra-distal radius in the LNG-IUS group at the seventh year of use and 0.456 ± 0.009 and 0.399 ± 0.010, respectively, at the 10th year of use. There were no significant differences between the two evaluations. In the LNG-IUS users who were experiencing some form of bleeding at the seventh year of use but had become amenorrhoeic by the 10th year, no statistically significant differences were found in BMD at the midshaft of the ulna or at the ultra-distal radius of the forearm (P < 0.341 and 0.983, respectively). The BMD (g/cm²) values obtained for users of the TCu380A IUD were similar to those of the LNG-IUS users, and there were no significant differences between the two groups at either evaluation time. In addition, at the seventh year of evaluation, the Z-score values for the ultra-distal radius were below −2SD in three LNG-IUS users (aged 41, 42 and 45 years, respectively), although at the 10th year, values were below this level in only one LNG-IUS user of 49 years of age. None of the IUD users had Z-scores below this value; however, there were no statistically significant differences between the two groups (Table II).

Table II

Bone mineral density (g/cm²) and Z-score according to the contraceptive method used, the section of the forearm evaluated and the moment of evaluation (n = 37 pairs).

Duration of use (months)	LNG-IUS¶ (n = 37)	TCu380A IUD@ (n = 37)	P-value
84 Months
Midshaft ulna	0.460 (0.009)	0.448 (0.008)	0.133•
Z-score less than −2.0 [n (%)]	4 (10.8)	1 (2.7)	0.358*
Ultra-distal radius	0.400 (0.011)	0.394 (0.006)	0.588•
Z-score less than −2.0 [n (%)]	3 (8.1)	0.0	0.240*
120 Months
Midshaft ulna	0.456 (0.009)	0.469 (0.008)	0.172•
Z-score less than −2.0 [n (%)]	3 (8.1)	0.0	0.240*
Ultra-distal radius	0.399 (0.010)	0.406 (0.009)	0.507•
Z-score less than −2.0 [n (%)]	1 (2.7)	0.0	>0.999*

^¶LNG-IUS, levonorgestrel-releasing intrauterine system; ^@IUD, intrauterine device; ^•Non-parametric Wilcoxon test for paired samples; *Fisher's exact test.

Table III shows the relationship between different variables and BMD at the 10th year of use in users of the LNG-IUS. With respect to age, 75% of the women were between 28 and 40 years of age. The number of participants with BMI (kg/m²) < 25 was almost the same as the number of women in whom BMI was ≥25. In women with higher BMI, BMD was higher at both sections of the forearm. Most of the women did not smoke and did not participate in any physical activities except for routine domestic chores. No statistically significant differences were found between the two groups with respect to any of the other characteristics.

Table III

Bone mineral density (g/cm²) at both sections of the forearm at the 10th year of LNG-IUS use according to the different variables.

Variables	n	Midshaft ulna (g/cm2)	P-value	Ultra-distal radius (g/cm2)	P-value
Age (years)			0.751¶		0.369¶
≤35	13	0.452 (0.012)		0.404 (0.017)
36–40	15	0.464 (0.012)		0.409 (0.013)
>40	9	0.450 (0.025)		0.374 (0.024)
BMI* (kg/m2)			<0.001+		0.002+
<25	19	0.426 (0.008)		0.371 (0.011)
≥25	18	0.489 (0.011)		0.428 (0.014)
Number of pregnancies			0.077+		0.102+
0–1	9	0.429 (0.016)		0.371 (0.018)
≥2	28	0.465 (0.010)		0.408 (0.011)
Smoker			0.275+		0.452+
Yes	3	0.424 (0.029)		0.374 (0.031)
No	34	0.459 (0.009)		0.401 (0.010)
Physical activity			0.496+		0.991+
Yes	4	0.474 (0.032)		0.399 (0.032)
No	33	0.454 (0.009)		0.399 (0.010)

*BMI, body mass index; ^¶Snedecor test (ANOVA); ⁺Student t-test for independent samples.

Multiple logistic regression analysis showed that higher BMD values at the midshaft of the ulna and the ultra-distal radius at the 10th year of evaluation were significantly associated with a higher BMI (kg/m²) and a higher BMD at the seventh year of evaluation and with a bleeding pattern of amenorrhoea (Table IV).

Table IV

Variables associated with bone mineral density in users of the LNG-IUS at the 10th year of use according to multiple linear regression analysis (n = 37).

Model/Variable	Coefficient	EP coefficient	P-value
Model 1: BMD at 10 years of use
BMD at the midshaft of the ulna at the seventh year of use (g/cm2)	0.798	0.050	<0.001
BMI (kg/m2)	0.003	0.001	0.003
Amenorrhoea	0.024	0.011	0.033
Model 2: BMD at the ultra-distal radius at 10 years of use
BMD at the ultra-distal radius at the seventh year of use (g/cm2)	0.663	0.075	<0.001
BMI (kg/m2)	0.005	0.001	<0.001

LNG-IUS, levonorgestrel-releasing intrauterine system; Dependent variable: bone mineral density (g/cm²; BMD; at the 10th year of use of the LNG-IUS); Model 1: BMD at the midshaft ulna; Model 2: BMD at the ultra-distal radius; Independent variables: BMD (midshaft ulna or ultra-distal radius) at the seventh year of use; age (in years); BMI (kg/m²); number of deliveries (<2/≥2); race (white/non-white); smoker (Yes/No); physical activity (Yes/No); occupation (physical/non-physical); age at menarche (years); bleeding pattern (amenorrhoea/bleeding); exclusive breastfeeding (<2/≥2 months); schooling (elementary/other); coffee drinker (yes/no); alcohol consumption (yes/no).

Discussion

In this study, forearm BMD of users of the LNG-IUS was evaluated at the seventh year of use, at which time the women received a new device. BMD was re-evaluated in these women at the 10th year of use and no significant differences were found between the two measurements. In addition, there were no significant differences between the BMD measurements of these women and those of a cohort of IUD users paired for duration of use, age, BMI (kg/m²), ethnicity and obstetric history. Moreover, Z-scores below −2SD (Consensus Development Conference, 1993; World Health Organization, 1994) at the ultra-distal radius were found in a few LNG-IUS users, all of whom were of menopausal age.

These findings were to be expected due to the fact that the most important variables related to bone loss and resorption in women is the hypoestrogenism (Bagger et al., 2004) observed during the post-menopausal years, exclusive breastfeeding and current use of DMPA (Guilbert et al., 2009). It was to be expected that the BMD of women in the LNG-IUS group would remain unaffected, since E₂ levels in users of this device are similar to those of women in the follicular phase of the menstrual cycle, as was found in the present study (Luukkainen and Toivonen, 1995; Bahamondes et al., 2003, 2006a). Moreover, values have been found to remain similar even when the women have been using the same device continuously for up to 9 years without renewal (Hidalgo et al., 2009), with values from 100 to 400 pg/ml.

Although BMD measurements were similar in both users and controls, there were some limitations to the present study, but also some strong-points. The main limitation is that LNG-IUS users were evaluated at the 7th and 10th years; however, baseline BMD measurements were not available, either for the users or the controls. Consequently, it is not possible to guarantee that the absence of any difference in BMD between the two evaluation times, and that there was no consequence of prolonged exposure to the LNG-IUS and/or to the several years of amenorrhoea. It could be argued that LNG-IUS users already lose BMD during the first years of use and by the 7th and 10th years of use, they may already be on the declining slope of bone loss, as occurs with DMPA users. However, there were no statistically significant differences in Z-score values between the groups.

BMD loss has been reported to be most profound in DMPA users in the first 1 or 2 years of use; however, this is a controversial issue. One study failed to find any differences according to duration of use (Tang et al., 2000), albeit another reported that mean BMD decreased as a function of the increase in the number of years of DMPA use, and that the rate of bone loss was higher in the first and second years of exposure (Scholes et al., 2002; Berenson et al., 2004). Nevertheless, DMPA provokes profound hypoestrogenism and the LNG-IUS does not. Moreover, the fact that there were no differences in BMD between the women in the two contraceptive study groups provides strength to our findings.

The effects of LNG and other 19-nor-progestogens on bone health remain controversial. LNG was considered to have a beneficial effect on bone turnover and BMD (Petitti et al., 2000) and was able to inhibit bone absorption in women of reproductive age (Volpe et al., 1997). In users of LNG-releasing implants (Norplant^® or Jadelle^®), no adverse effects on BMD have been found at the lumbar spine (L2–L4), femoral neck nor ultra-distal radius (Naessen et al., 1995; Petitti et al., 2000; Bahamondes et al., 2006b). In addition, it has been established (Enríquez et al., 2007) that the 3β,5α-tetrahydro reduced derivatives of LNG exert a significant effect on calvarial osteoblasts from newborn rats, and stimulate osteoblast proliferation, differentiation and mineralization processes, mimicking the effects of E₂ and resembling the potency observed with E_2. Although this effect is in agreement with the concept that LNG is an effective agent in bone remodeling, it was observed only in post-menopausal women (Liu and Muse, 2005).

In addition, it could be argued that BMD was not evaluated at the lumbar spine or trochanter. Nevertheless, it has been established that forearm BMD constitutes a good indicator of bone health (Marshall et al., 1996). Jones and Davie (1998) evaluated forearm BMD in women to identify osteoporosis at the spine or femoral neck and reported that this method successfully detected the condition in 62% of subjects with osteoporosis (at the spine, femoral neck or at both sites), with a specificity of 89%. The need to treat osteoporosis was correctly defined in 90% of cases. In addition, Picard et al. (2004) reported good sensitivity (83%) and specificity (84%) with forearm BMD when a combination of two forearm measurements was taken into account and that forearm BMD may represent an interesting tool for the diagnosis of osteoporosis.

Another limitation is the fact that by the 10th year of evaluation, the IUD users group had had significantly more pregnancies (2.4) compared with the LNG-IUS users (2.1) and this may also have influenced the fact that BMD was similar in both users and non-users. The influence of the number of pregnancies on BMD is also a controversial issue, particularly many years after the end of the last pregnancy, as occurred in the present cohort of women, whose last pregnancy took place at least 10 years prior to entering the study. Gur et al. (2003) reported a significant correlation between the number of pregnancies and BMD at the spine, trochanter and Ward's triangle in post-menopausal women who had had more than five pregnancies; whereas, no differences were observed either in nullipara or in women who had had one or two pregnancies. Others (Paton et al., 2003) measured BMD in adult women and found no significant differences according to the number of pregnancies. However, women with 1, 2 or ≥3 pregnancies had higher adjusted lumbar spine BMD compared with nulliparous women. Therefore, these authors concluded that pregnancy did not exert any long-term detrimental effect on BMD.

Almost 50 and 90% of LNG-IUS users reported amenorrhoea at the 7th and 10th years of use, respectively, and these figures are similar to previous findings (Ronnerdag and Odlind, 1999; Hidalgo et al., 2002). This bleeding pattern may lead users, potential users and physicians to believe that the device could affect E₂ levels and, consequently, BMD. Considering that the LNG-IUS is also used for non-contraceptive purposes (Luukkainen and Toivonen, 1995; Bahamondes et al., 2008) and is an appropriate method for women who need to avoid the use of estrogen, it is important to show that there is no effect on ovarian E₂ production or BMD. Amenorrhoea is a result of the high local tissue concentrations of LNG and the antiproliferative action of the steroid in the endometrium. In addition, anovulation has been reported in some LNG-IUS users with increased E₂ production, and higher plasma E₂ levels have been found in amenorrhoeic users compared with menstruating women (Luukkainen and Toivonen, 1995). Consequently, the present E₂ and BMD results in amenorrhoeic women were to be expected.

It could be argued that BMD was similar in LNG-IUS and IUD users because there were more smokers in the latter group and this fact could influence the results due to the known effect of smoking on BMD (Tamaki et al., 2009). These authors showed that female ever-smokers (aged 20–40 years) were significantly associated with decreased lumbar BMD following adjustment for age, height and weight; however, BMD was lower in women who smoked three pack-years of tobacco, a finding not observed in the present sample.

Many factors may also affect peak bone mass acquisition. In addition to genetic factors, other variables affecting bone mass include BMI, diet, calcium intake, physical activity and hormone status (Sarfati and de Vernejoul, 2009). Dietary calcium intake was not controlled in the present study. Nevertheless, in the group of LNG-IUS users, other characteristics such as exclusive breastfeeding, physical activity and alcohol and coffee consumption did not appear to affect BMD results. In addition, coffee consumption appears to affect only DMPA users, with no effect having been found in other groups of women (Wetmore et al., 2008).

This is the second report from our group (Bahamondes et al., 2006a) on the effect of the LNG-IUS on the BMD of women using the device for contraceptive purposes and the first report to evaluate women using this method for a period of up to 10 years that included removal of the initial LNG-IUS and immediate placement of a second device. This fact should be considered the principal strength of the study. Nevertheless, there are other limitations that should be taken into account. The study population consists predominantly of young white women. These women were, in general, neither close to the menopause nor to adolescence, the age-groups in which the effect of progestogen-only contraceptives could have a greater effect on BMD. Bone loss is predicted by the difference between bone formation and bone resorption, and currently available markers, including BMD, are insufficient to enable this balance to be calculated. However, it is reasonable to conclude that forearm BMD at the midshaft of the ulna and ultra-distal radius remained unchanged between the 7th and the 10th years of use of the LNG-IUS and similar to that of IUD users.

Funding

Partial financial support was received from the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) under grant number 03/083917 and also from the Conselho Nacional de Pesquisa (CNPq) under grant number 573747/2008-3.