Levonorgestrel Intrauterine Device Use in Overweight and Obese Women

Abstract

The levonorgestrel intrauterine device (LNG-IUD) is a safe, effective, long-acting, reversible contraceptive that reduces unintended pregnancy and decreases heavy menstrual bleeding. Many procedures such as IUD insertion are more challenging in overweight and obese women. The objective of this study was to describe LNG-IUD insertion, continuation, and complications in overweight and obese women in an ethnically diverse population in Hawai‘i. A retrospective cohort study of women who had a LNG-IUD inserted at the University of Hawai‘i, Department of Obstetrics and Gynecology Resident and Faculty practice sites between January 2009 and December 2010 was performed. A total of 149 women were followed. The most commonly reported races were Asian (32%), Native Hawaiian (26%), and non-Hawaiian Pacific Islander (20%). The mean BMI of the study population was 28.4 (standard deviation 7.2) with 37% classified as normal weight, 30% as overweight, and 33% as obese. Overall, 76% of women continued the LNG-IUD 12 months after insertion. No statistically significant difference emerged in 12-month IUD continuation between the BMI groups. Difficult (5%) and failed (3%) IUD insertions were rare for all BMI groups. IUD complications occurred in 9% of women and included expulsion and self-removal. In this diverse population, the majority of women continued to use the LNG-IUD one year after insertion with low rates of difficult insertions and complications.

Introduction

Among women of reproductive age in the United States, 59.5% are considered overweight or obese.¹ In Hawai‘i, approximately 46% of adult women 18 years and older are overweight or obese.² Overweight and obese women have higher rates of menstrual irregularity and complications related to pregnancy.³ Studies suggest rates of unintended pregnancy may vary by body mass index (BMI) with rates being higher in obese women.⁴ Though unwanted pregnancy is an important health concern for all women, the health risks to obese women may be greater.⁵

Providing contraception to obese and overweight women to help them achieve their reproductive life goals raises specific challenges. Combined hormonal contraceptive methods like the pill may be less effective.^6,7 The contraceptive patch has been shown to be less effective in women weighing more than 90 kilograms (198 pounds).⁸ Although venous thromboembolism is rare in reproductive age women, studies suggest obese women taking combined hormonal contraceptives have an increased risk of venous thromboembolism compared to normal weight women.⁹ Compliance with an oral contraceptive pill differed by body weight in the research setting with obese women being less likely to adhere to the study regimen than normal weight women.¹⁰

The intrauterine device (IUD) is safe and highly effective regardless of body weight due to its direct progestogenic effect on the endometrium, making it an ideal method for many women.¹¹ IUDs are easy to use, have few absolute contraindications, and result in few side effects beyond changes in menstrual bleeding. The levonorgestrel IUD (LNG-IUD) results in decreased menstrual bleeding which many women find desirable.¹²

Cochrane Reviews in 2010 and 2013 of hormonal contraceptive use in overweight or obese women found no studies describing hormonal IUD use in obese women.^13,14 Subsequent to the 2010 Cochrane Review, three studies evaluating IUD use in obese women have been published. One study examined acceptance rates of IUDs in obese adolescents,¹⁵ and two studies looked at choice of contraception in women of different BMIs.^16,17 None of these studies evaluated IUD insertion, complications, or continuation rates. Given that insertion of IUDs may be more challenging in overweight and obese women, studies evaluating these outcomes are needed to understand whether IUDs are an effective form of contraception for this population. Additionally, data on IUD use in a significant number of Native Hawaiian and Pacific Islander women has not been previously published. The objective of this study was to describe LNG-IUD use in overweight and obese women based on different BMI categories. Specific measures included difficulties with IUD insertion, complications, and 12-month continuation rates.

Methods

Study Population

We conducted a retrospective cohort study of female patients of all ages who had an IUD-related medical visit at the University of Hawai‘i (UH), Department of Obstetrics and Gynecology resident and faculty practice sites between January 1, 2009 and December 31, 2010. Eligible patients were identified through the electronic medical record using International Classification of Disease, 9^th revision (ICD-9) codes for IUD use [V25.1, V25.11, V25.12, V25.13, V25.42, 69.7, 97.71] and IUD complications [996.32, 996.65, 996.76] and Current Procedural Terminology (CPT) codes for IUD insertion and removal [58300, 58301]. Our study population was further refined to patients who had a LNG-IUD inserted at a UH resident or faculty practice site and had a follow-up visit documented in the medical record more than 365 days after the insertion date. During the study period, only the LNG20-IUD (releases 20 mcg LNG per day, Mirena®, Bayer HealthCare Pharmaceuticals Inc.) was available. Therefore, the LNG14-IUD (releases 14 mcg LNG per day, Skyla®, Bayer HealthCare Pharmaceuticals Inc.) was not included in this study. Two authors [LST, SH] reviewed all of the medical records. For quality purposes, data were compared between the two chart reviewers.

Main Independent Variable

The main independent variable was body mass index (kg/m²) at the time of IUD insertion. The patient's measured weight on the day of IUD insertion and the measured or self-reported height were used to calculate the BMI. It was decided a priori to evaluate BMI as a dichotomous variable (normal weight [BMI < 24.9] vs overweight and obese [BMI ≥ 25]). During data analysis, clinical differences between the overweight and obese groups became apparent. Therefore, the results are presented as three BMI categories (normal weight [BMI < 24.9], overweight [BMI 25–29.9], and obese [BMI ≥ 30]).¹⁸

Dependent Variables

Dependent variables including successful IUD insertion, difficult IUD insertion, 12-month continuation, and complications related to the LNG-IUD. IUD insertion were evaluated based on the IUD insertion procedure note. Successful IUD insertion was defined as placement of the LNG-IUD at the first IUD insertion visit without use of additional instrumentation (ie, cervical dilators), anesthesia, or cervical ripening agents. Difficulty with IUD insertion (yes/no) was defined as successful placement of the IUD at the initial visit, but with the use of additional instrumentation, anesthesia or cervical ripening agents. Failed IUD insertion was defined as the inability to insert the IUD on the day of the procedure.

The 12-month continuation rates and complications of the LNG-IUD were determined by reviewing all the office or emergency room visit notes in the electronic medical record after the IUD was inserted. The 12-month continuation of the IUD was categorized as “yes” if the IUD was still in place 12 months or more after insertion and “no” otherwise. The IUD was determined to be in the uterus by visualization of the IUD strings through the external cervical os or any diagnostic imaging report describing an IUD in the uterus. Once a patient had her IUD placed, it was assumed that the same IUD remained in situ unless the patient reported removal and/or replacement of her IUD at an outside facility.

Complications from the LNG-IUD included infection, expulsion, perforation, or failure. Infection was defined as signs or symptoms of pelvic inflammatory disease after the IUD was placed. Expulsion of the IUD was defined as the patient or provider reporting visualization of the IUD outside the uterus or if no strings were visualized through the cervical os, an ultrasound confirmed absence of the IUD in the uterus and an x-ray confirmed absence of the device in the abdominopelvic cavity. Perforation was defined as imaging reports with the IUD through the myometrium or in the abdominal cavity. IUD failure was defined as a diagnosed pregnancy with the IUD in situ.

Statistical Analysis

Bivariate analysis was used to evaluate the relationship between the main independent variable and the outcomes using the Chi-square test for categorical variables and the Student's t-test for continuous variables. Data on potential confounders were collected including age, marital status, insurance type, ethnicity, race (the electronic medical record allows patients to identify with only one race, obtained verbally during registration at the clinic or hospital), gravity and parity, history of abortion, history of sexually transmitted infections (STIs) or pelvic inflammatory disease, reason for IUD insertion, training level of inserting provider, postpartum insertion, and breastfeeding. A multiple logistic regression model was created with body mass index and any potential confounders and predictor variables that were associated with the outcome using a P <.20 cutoff for inclusion in the model. Through backward elimination, confounders or predictor variables were removed from the model at a significance of P >.05.

A sample size calculation was used to determine the number of charts to review based on the assumption that 85% of normal weight women would be using the LNG-IUD after 12-months.^19,20 To detect a 10 percent difference in 12-month continuation between normal weight versus overweight and obese women with 80% power and a one-sided alpha of 0.05, a total of 280 charts were to be reviewed (140 charts in the normal weight group and 140 charts in the obese and overweight group).

The data was analyzed using SPSS 16.0 (SPSS Inc., Chicago, IL). The statistical significance was set at P <.05. The Western Institutional Review Board approved this study.

Results

Of the 416 potentially eligible patients identified by ICD-9 and CPT codes, 149 met our inclusion criteria (Figure 1). Most of the patients were between 21 and 30 years old (47%), single (69%), and used public insurance (61%) (Table 1). The most commonly reported races were Asian (32%), Native Hawaiian (26%), and non-Hawaiian Pacific Islander (20%). The majority of the patients had given birth to at least one child (88%), did not have a previous abortion (70%) or a history of sexually transmitted infection (75%) or pelvic inflammatory disease (98%). Most women used the LNG-IUD for contraception (91%) rather than for menstrual irregularity (9%). Resident physicians inserted seventy-five percent of the LNG-IUDs. Of the women who had the LNG-IUD inserted postpartum, 78% had the LNG-IUD placed more than 6 weeks after delivery and 72% of these women were breastfeeding.

Figure 1.

Patient Eligibility

Table 1.

Patient characteristics of women with LNG-IUDs by BMI category

	Body Mass Index Category
Characteristics	Normal weight n=55	Overweight n=45	Obese n=49	Total N=149	P-valuea
Age					.10
≤20	19 (35)	7 (16)	7 (14)	33 (22)
21–30	22 (40)	23 (51)	25 (51)	70 (47)
>30	14 (25)	15 (33)	17 (35)	46 (31)
Marital Status					.06
Single	40 (75)	25 (56)	37 (76)	102 (69)
Married	13 (25)	18 (40)	11 (22)	42 (29)
Divorced	0 (0)	2 (4)	0 (0)	2 (1)
Widowed	0 (0)	0 (0)	1(2)	1(1)
Insurance					.02
Public	30 (55)	22 (49)	39 (80)	91 (61)
Private	19 (35)	19 (42)	8 (16)	46 (31)
No Insurance	6 (11)	4 (9)	2 (4)	12 (8)
Race					<.01
African American	3 (6)	1 (3)	0 (0)	4 (3)
American Indian	0 (0)	0 (0)	3 (7)	3 (2)
Asian	24 (47)	13 (33)	6 (14)	43 (32)
Caucasian	10 (20)	5 (13)	2 (5)	17 (13)
Hispanic	3 (6)	1 (3)	1 (2)	5 (4)
Native Hawaiian	10 (20)	11 (28)	14 (33)	35 (26)
Pacific Islander	1 (2)	9 (23)	16 (38)	26 (20)
Parity					.10
0	9 (16)	2 (4)	6 (12)	17 (11)
1–2	32 (58)	30 (67)	22 (45)	84 (56)
3 or more	14 (26)	13 (29)	21 (43)	48 (32)
Previous Abortion					.63
Yes	18 (33)	14 (31)	12 (24)	44 (30)
No	37 (67)	31 (69)	37 (76)	105 (70)
Previous Sexually Transmitted Infection					.35
Yes	14 (25)	8 (18)	15 (31)	37 (25)
No	41 (75)	37 (82)	34 (69)	112 (75)
Previous Pelvic Inflammatory Disease					.42
Yes	2 (4)	1 (2)	0 (0)	3 (2)
No	53 (96)	44 (98)	49 (100)	146 (98)
Reason for IUD Insertion					.29
Contraception	53 (96)	41 (91)	42 (86)	136 (91)
Menstrual Irregularity	1 (2)	1 (2)	4 (8)	6 (4)
Both	1 (2)	3 (7)	3 (6)	7 (5)
Inserting Provider					<.01
Resident	33 (60)	33 (73)	46 (94)	112 (75)
Attending	16 (29)	10 (22)	3 (6)	29 (20)
Nurse Practitioner	6 (11)	2 (4)	0 (0)	8 (5)
More than 6 weeks postpartum					.82
Yes	28 (82)	21 (78)	24 (73)	73 (78)
No	6 (18)	6 (22)	9 (27)	21 (22)
Breastfeeding					.58
Yes	18 (67)	15 (79)	20 (71)	53 (72)
No	9 (33)	4 (21)	8 (29)	21 (28)

^aP-values were determined by Chi-square test. Column percentages may not add up to 100 percent due to rounding.

The mean body mass index for the study population was 28.4 kg/m² (SD 7.2) with 37% (n = 55) classified as normal weight, 30% (n = 45) as overweight, and 33% (n = 49) as obese. BMI varied significantly by race (P <.01) (Table 1). A higher proportion of Asian (24/43, or 56%) and Caucasian (10/17, or 59%) women were of normal BMI, while Native Hawaiian (25/35, or 71%) and non-Hawaiian Pacific Islanders (25/26, or 96%) were more likely to be overweight or obese (P <.01). Only one non-Hawaiian Pacific Islander woman had a normal BMI. BMI also varied by level of inserting provider as resident physicians were more likely to insert LNG-IUDs in the overweight and obese women, while attending physicians and nurse practitioners inserted more of the LNG-IUDs in the normal weight women (P <.01).

Most LNG-IUDs (92%) were inserted without difficulty (Table 2). Difficulty with IUD insertion, though not statistically significant, was more common in the normal weight (5%) and obese groups (8%) than the overweight group (2%; P =.47). Failed IUD insertion occurred in two patients in the normal weight group secondary to cervical stenosis and two patients in the obese group, one due to cervical stenosis and the other due to inability to visualize the cervix.

Table 2.

Twelve-month LNG-IUD continuation, Difficult with IUD insertion, and Complications by BMI category

	Normal weight	Overweight	Obese	Total	P-valuea
Difficulty with IUD Insertion					.47
Yes	3 (5)	1 (2)	4 (8)	8 (5)
No	50 (91)	44 (98)	43 (88)	137 (92)
Failed Insertion	2 (4)	0 (0)	2 (4)	4 (3)
IUD in place at 12 months					.73
Yes	39 (74)	36 (80)	35 (74)	110 (76)
No	14 (26)	9 (20)	12 (26)	35 (24)
Complications					.47
Yes	6 (11)	2 (4)	5 (11)	13 (9)
No	47 (89)	43 (96)	42 (89)	132 (91)

^aP-values were determined by Chi-square test.

The majority of patients (76%) had their IUD in place 12 months after insertion (Table 2). No statistically significant difference emerged in 12-month IUD continuation between the BMI groups (P =.73). Complications were higher in the normal weight (11%) and obese (11%) groups compared to the overweight group (4%), but were not statistically significant (P =.47). Reported complications included expulsion (11/145, 8%) and self-removal (2/145, 1%), patients deliberately and successfully removing their own LNG-IUD. Both self-removals occurred in normal weight women. No perforations, infections, or failures occurred in the study population.

In the final multiple logistic regression model, adjusting for age and insurance, overweight women had a 33% higher odds (adjusted OR 1.33, 95% CI 0.48-3.66) of continuing the LNG-IUD 12 months after insertion compared with normal weight women, while obese women had a 25% lower odds (adjusted OR 0.75, 95% CI 0.29–1.92) of continuing the LNG-IUD for 12 months after insertion compared to normal weight women (Table 3).

Table 3.

Unadjusted and Adjusted OR for 12-month LNG-IUD continuation rates (OR, 95% CI)

	Unadjusted OR	Adjusted OR*
BMI Category
Normal weight	Reference	Reference
Overweight	1.34 (0.43–2.57)	1.33 (0.48–3.66)
Obese	0.73 (0.27–1.95)	0.75 (0.29–1.92)

^*Adjusted for Age and Insurance. Other confounders (eg, race) were not included in the model because the P-value was ≥.20 in bivariate analysis.

Discussion

A higher than expected loss to follow up resulted in this study being underpowered to demonstrate statistically significant differences in IUD continuation based on BMI. However, we provide estimates of 12-month IUD continuation by BMI group, which ranged between 74% and 80% and is similar to general continuation rates that have been reported in the literature (70% to 88%).^19–21 Difficult (2% to 8%) and failed (0% to 4%) IUD insertions were rare for all groups and were similar to rates reported in a recent study of LNG-IUD and Copper IUD insertion in nulliparous women, which classified 9.3% as difficult and 1.2% as failed IUD insertions.²²

IUD expulsion and self-removal occurred in a minority of women. The overall IUD expulsion rate of 8% was higher than expected when compared to 3 to 5% which is typically reported in the medical literature.^19,21 However, a study by Teal et al. of IUD use in primiparous teenagers reported a LNG-IUD expulsion rate of 13.3%, likely due to the younger patient age.²³ Additionally, Madden, et al, noted a 12-month expulsion rate of 6.3 per 100 LNG-IUDs and a 36-month cumulative expulsion rate of 10.1 per 100 LNG-IUDs.²⁴ Larger studies of an ethnically diverse population with adequate power are needed to better understand our findings.

As noted earlier, half of the women did not follow up 12-months after IUD insertion. Lower than expected follow up rates may be reflective of patients who are satisfied with a long-term contraceptive and do not need to return for birth control refills. Additionally, decreases in the recommended frequency of cervical cancer screening may have contributed to lower than expected follow up.

Despite our racially diverse population, we used the National Institute of Health (NIH) standard BMI classification to define obesity. Race-specific BMI cutoff points have been proposed for the Asian population in general, but have not been determined for each subset of Asians (ie, Japanese, Chinese, Korean).²⁵ Since we do not have race-specific BMI cutoffs for each race in our population, the NIH BMI classification was used for consistency among the population. Once race-specific BMI cutoffs are defined, it would be useful to evaluate IUD use by these BMI categories.

BMI misclassification is a potential limitation as self-reported height was used to calculate the patient's BMI if a measured height was not available. Studies have documented that self-reported height and weight provide a reasonable representation of women's BMIs, particularly when BMI is categorized.²⁶ We used BMI on the day of LNG-IUD insertion because, unlike BMIs measured on other days, this BMI could affect the success of IUD insertion. Once the LNG-IUD is inserted into the uterus, a change in the patient's weight does not affect the LNG-IUD's mechanism of action and therefore would unlikely affect continuation and complications.

The study population was racially diverse with the majority of patients being Asian, Pacific Islander, and Native Hawaiian, which is reflective of Hawai‘i's demographics. Although not exactly comparable to our study, Thompson, et al, described the demographic characteristics and the proportion of IUD users in a diverse population in California.²⁷ They found that a smaller proportion of US-born Asian women, compared to other ethnic groups, used intrauterine contraception.

Overall, this is the first known study to evaluate LNG-IUD use by BMI in Asian and Pacific Islander women. Additional studies are needed with a larger baseline population and less loss to follow up. A larger sample will allow for stratification by classes of obesity and other demographic variables and will provide more information on this understudied research topic. Also, to adequately understand IUD effectiveness in preventing unintended pregnancy, we need to collect information on important potential cofounders such as frequency of sexual intercourse, use of alternative forms of birth control (eg, condoms, female condoms, dental dams, male sterilization), and underlying infertility or other medical conditions. Nevertheless, this study provides estimates of IUD continuation and complications which may inform future research and guide recommendations for its use in the clinical setting.

Conflict of Interest

None of the authors identify a conflict of interest.