Progestin-only contraceptive pill use among women in the United States.

Oral contraceptives (OCs) are used by over 10 million U.S. women [ 1 ]. Combined OCs (COC), which contain both estrogen and progestin, are highly effective and have an excellent safety profile [ 1 , 2 ]. However, estrogen-containing formulations have long been known to increase the risk of venous thromboembolic events (VTE) [ 3 - 7 ]. Estrogen dosages in contraceptive products have decreased dramatically over the last five decades and most OC users have few risk factors and low absolute risk of VTE [ 2 ]. Yet, women using COCs have a 2- to 6-fold greater relative risk of VTE compared to non-users [ 2 , 8 ]. For women with higher risk of VTE, even combined hormonal contraceptives with lower dosages of estrogen may be clinically relevant [ 2 - 7 ]. In particular, obesity, diabetes, and hypertension all increasingly affect U.S. women of reproductive-age [ 9 - 13 ]; pregnancy and tobacco use are additional risk factors for VTE [ 2 ]. For women at increased risk of VTE, progestin-only contraceptive methods have particular advantages [ 2 - 4 ]. Progestin-only pills (POPs) are believed to have similar effectiveness to COCs [ 2 ] and do not increase the risk of VTE [ 8 ], yet use of POPs by U.S. women has not been documented. Using the most recent population-based data, we estimated the prevalence of POP use among U.S. women, including women with VTE risk factors.

Our sample included all women ages 15 to 44 years (n=12,279) who participated in the most recent National Survey of Family Growth (NSFG), a nationally representative survey conducted by the National Center for Health Statistics. The U.S. population-based survey collects information on family life, marriage and divorce, pregnancy, infertility, use of contraception, and men and women's health. Data were collected continuously between 2006 and 2010 via household, in-person interviews. African American and Hispanic women and adolescents and young adults 15 to 24 years were oversampled. The survey response rate was 77%. Further information about the design and sampling of the NSFG can be found at http://cdc.govnchs/nsfg.htm [ 14 ]. The Institutional Review Boards of Princeton University and the University of Pittsburgh approved this study. We focused on a series of questions regarding contraceptive use and specifically OC use. Women were first asked during the interview whether they had ever used a contraceptive method. They were further asked the type of contraceptive method they used during the month of the interview as well as in the month prior to the interview. Women who reported having used OCs in the current or previous month were presented with a chart of 98 OCs commonly used in the U.S. and asked to identify the brand of pill from the chart that they currently or most recently used. If they were not able to identify the pill from the chart, they were asked to present their pill pack to the interviewer. The list included six available progestin-only pills (all of which contain 0.35 mg norethindrone): Camilla®, Errin® Micronor®, Nor-QD®, Jolivette® and Nora-BE®. We considered women who identified any of these six formulations as their method and who reported using the pill in the month they were interviewed to be current POP users. We used descriptive statistics to describe the sample and provide estimates of the proportion of women using POPs overall and by type of POP. We provide a population estimate of POP use based upon the 2010 data from the U.S. Census Bureau's estimate of the population of women ages 15-44 years (N=62,374,964) [ 15 ]. We also estimated the proportion of women using COCs who are at risk of an unintended pregnancy (non-pregnant, non-sterile, sexually active women not trying to get pregnant) [ 1 ]. We further describe and compare the proportions of women currently using POPs in our sample across sociodemographic groups and by key health-related and reproductive characteristics using bivariate chi-square tests. We also examine POP use among women with VTE risk factors according to the WHO and CDC medical eligibility criteria for hormonal contraceptive use [ 3 , 4 ]. Risk factors assessed by the NSFG included age ≥35 years, overweight (25-29.9kg/m 2 ) or obese (≥30 kg/m 2 ) body mass index (BMI) (versus underweight <18.5 kg/m 2 or normal weight 18.5-24.9 kg/m 2 BMIs), non-gestational diabetes, cigarette smoking a pack per day or more (versus none or less than a pack per day in the past year), and postpartum status (having delivered within the three months prior to the interview). Each of these risk factors alone do not restrict use of COCs (i.e., they are eligibility categories 1 or 2) [ 3 , 4 ]. However, in combination or in the setting of other VTE risk factors not assessed by the NSFG (e.g., hypertension or hyperlipidemia), these conditions preclude use of COCs (i.e., they are rated categories 3 or 4) [ 3 , 4 ]. Due to small sub-sample sizes, we were unable to examine POP use among women with multiple risk factors. We report unweighted frequencies (n), but in all analyses weighted data were used to account for the complex, stratified sampling design of the survey; weighted proportions (%) and tests of significance were computed using the svy series of commands in Stata 11.0 (Stata Corporation, College Station, TX).

Sample characteristics are presented in Table 1 . The mean (± standard deviation) age of participants (n=12,279) was 29 ± 8 years. The majority of women identified their race/ethnicity as white (62%). Over half the sample (52%) reported some college education. Nearly half of women (48%) resided in a sub-urban area; fewer in urban (32%) and rural (20%) communities. Forty-six percent of women reported incomes below 200% of the federal poverty level, although the majority (66%) were employed. Twenty-seven percent of women were uninsured at some point during the previous year. Over half had ever been married (53%) or cohabitated with a non-marital partner (53%). Most women (70%) reported one current sexual partner. Nearly two–thirds (61%) of women reported a history of pregnancy and 55% had given birth to one or more children. Few women reported being pregnant at the time of their interview (4%) or having delivered in the prior three months (2%); 2% were currently breastfeeding. A history of a gynecological problem was reported by 30% of women, including problems with ovulation (17%), ovarian cysts (16%), uterine fibroids (6%) and endometriosis (5%). A significant number of women had risk factors for VTE. Thirty-four percent of women were aged 35 years or older. Nearly a quarter (23%) reported being overweight (BMI 25-29.9) and 28% reported being obese (BMI ≥30). Tobacco use was reported by 26% of women; 8% smoked a pack or more per day. Only 2% of women reported a history of non-gestational diabetes; 4% reported gestational diabetes. Eighty-eight percent of women reported ever having used a contraceptive method and 73% reported having used OCs. Among the women with a contraceptive history (n=10,779), 20% (n=2,032) (18% of all women) reported OC use in the month of the interview and 21% (n=2,072) reported OC use in the prior month. Of women currently using an OC (n=2,032), approximately 2% (n=57) identified a POP as their current contraceptive method ( Table 2 ). Overall, 0.4% of all women surveyed (n=12,279) were current POP users. This, along with the 2010 Census Bureau's population estimate that there are 62,374,964 women ages 15-44 years in the United States, suggests that approximately 249,500 US women currently use POPs. We further examined POP use among women at potential risk of pregnancy. When excluding pregnant women (n=516) or women who desire pregnancy and are therefore not using contraception (n=395), women with a history of surgical (n=1,857) or non-surgical (n=203) sterility, those with no history of sexual intercourse (n=1,674) or no intercourse in the last 3 months (n=2,240), 0.7% of all women at risk of unintended pregnancy used POPs as their contraceptive method (n=53/7,660). The proportions of women using POPs according to pill type are shown in Table 2 . Among OC users (n=2,032), Micronor® was the most common POP used (0.8%), followed by Errin® (0.8%), Camilla® (0.4%), Nor-QD® (0.2%), Jolivette® (0.2%) and Nora-BE® (0.1%). POP use was similar across sociodemographic groups ( Table 3 ). POP use did not vary by reason for OC use (p=0.68) nor by the type of medical practice from which a woman had received birth control within the previous 12 months (p=0.68). However, POP use varied by location of residence (p=0.003); among OC users, rural women had higher proportions of POP use than urban and sub-urban women (5% rural versus 1% urban and 2% suburban). Women who had been married were more commonly POP users than never married women (4% versus 1%, p<0.001). POP use varied by reproductive characteristics ( Table 3 ). POP varied by parity (p1 childbirths, 4% for 1 childbirth versus 1% for nulliparous) and women with a late age at menarche (6% for age 14 years and older versus 2% for 11-14 years and 3% for <11 years). POP use was higher among those with a history of ovarian cysts (4% versus 2%, p=0.03) and among postpartum women (32% versus 2%, p<0.001) as compared to their counterparts. Finally, POP use by VTE risk factors is presented in Table 4 . POP use varied by intensity of smoking (p=0.008); among women using OC, those who smoked a pack of cigarettes or more per day (5%) had higher POP use than women who smoked less than a pack per day (0.2%) or who did not smoke (3%). POP use was similar among women <35 versus ≥35 years of age (p=0.22), across BMI groups (p=0.37) and with or without a diagnosis of non-gestational diabetes (p=0.65).

This analysis of nationally-representative U.S. data indicates that despite the large number of US women using OCs, very few are currently using POPs. Most commonly, POPs are used in the postpartum period. Given women's higher risk of VTE immediately postpartum and the potential effects of estrogen on breast milk supply and infant development, it is prudent to avoid estrogen during this time [ 3 , 4 , 16 , 17 ]. In addition, women who smoke had higher proportions of POP use than nonsmokers. Though, women with other risk factors for VTE did not have higher proportions of POP use. Use of POPs among women with higher body weights and older age was similar to that of thinner and younger women. None of the OC-using women with a diagnosis of diabetes were using POPs. The American College of Obstetricians and Gynecologists (ACOG) suggests that obesity and the use of COCs are independent risk factors for venous thromboembolism and that obese women should avoid estrogen-containing contraceptives when possible alternative contraceptive options exist because of the increased risk [ 16 ]. However, other organizations have issued somewhat divergent guidelines, allowing the use of estrogen-containing contraceptives by obese women [ 3 , 18 ]. Similarly, although guidelines recommend that diabetic women with end organ damage avoid estrogen, use of estrogen by women who have had diabetes for only a short period of time is considered acceptable [ 3 , 4 ] Additionally, many clinicians remain concerned that POPs are less tolerated and therefore less effective than estrogen-containing contraceptives [ 19 ]. However, studies have reported similar satisfaction and continuation rates for women who used POPs or COCs [ 19 - 21 ]. Recommendations for specific use of POPs, such as stringent daily timing (POPs should be taken at the same time daily) and missed or late pills rules (back-up contraception is recommended for pills taken more than 3 h late), may also play an important role in clinicians perceptions of POPs and ultimately hinder their provision [ 22 ]. In Northern Europe and other regions, availability of POP formulations which more consistently suppress ovulation and contribute to less irregular bleeding (such as 0.075mg desogestrel) than formulations available in the U.S., may help explain their more widespread use in these settings [ 23 , 24 ]. The inconsistency around opinions and practices of prescribing and using POPs as well as women's access to different formulations highlights the need for further study of the comparative acceptability and effectiveness of POPs and COCs for women in the U.S. and abroad. This study has a number of limitations, in particular, the lack of data on end-organ disease among diabetics, hypertension, hyperlipidemia or other relative contraindications to estrogen use. More broadly, these data do not make clear the full context in which women were choosing to use COC and/or POPs. The NSFG distinguishes pill type only with a single question about the OC currently used. Thus, we were unable to examine women's past use of POPs or COCs separately, pill switching, or POP dissatisfaction, which could have provided insight into the circumstances in which women chose or discontinued POPs within their contraceptive histories. It is possible that we were unable to fully disentangle relationships between BMI and race/ethnic characteristics, which may interact with age at menarche and other characteristics to contribute different rates of POP use [ 17 , 25 ]. Unfortunately, our sample of POP users was limited, precluding examination of the effects of multiple risk factors for VTE, multivariate analysis and our ability to identify the true determinants of POP use. In conclusion, very few women in the U.S. use POPs. Further study of the basis for OC selection is needed, especially for women with chronic conditions that increase risk of VTE. A better understanding of situational factors and contraceptive experiences that contribute to women's use or nonuse of POPs may help identify areas for counseling and intervention to promote POPs for women who prefer OCs to more highly effective reversible contraceptive methods such as intrauterine devices and implants.