Types of combined oral contraceptives used by US women.

The oral contraceptive pill (OC) is among the most favored contraceptive methods, used by 9% of women aged 15–49 years worldwide and by over 18% of reproductive-aged women in developed regions [ 1 ]. More U.S. women use COCs than any other method [ 2 , 3 ]. Over the last five decades, COCs have undergone significant developments [ 3 – 9 ]. Lower estrogen doses, modified progestin components, multiphasic formulations, and extended cycle regimens and shortened hormone-free intervals have been attempts to improve safety and tolerability profiles while maintaining efficacy [ 3 – 14 ]. The wide range of COCs formulations currently available may have implications for unintended pregnancy and adverse event rates but also for other factors important to modern women’s lives such as non-contraceptive health conditions and reproductive health care access (e.g., contraceptive insurance coverage) [ 5 – 9 , 14 – 18 ]. However, the types of pills women are using at the population level have not been recently described. We estimated the prevalence of COC use by type of COC and examined factors associated with type of COC use among women in the United States, 2006 to 2010.

We used the most recent data from The National Survey of Family Growth (NSFG), a nationally representative survey conducted periodically by the National Center for Health Statistics since 1973. The population-based survey collects information on family life, marriage and divorce, pregnancy, infertility, use of contraception, and men and women’s health. Between 2006 and 2010 (continuous cycle), in-person interviews were conducted in the homes of 22,682 U.S. women and men ages 15 to 44 years. African Americans and Hispanics and adolescents and young adults aged 15 to 24 years were oversampled. The response rate was 77%. Further information about the design and sampling of the NSFG can be found at http://cdc.govnchs/nsfg.htm [ 19 ]. Our sample included all women ages 15 to 44 years (n=12,279). The Institutional Review Board of Princeton University approved this study. NSFG questions assessed lifetime contraceptive use and the contraceptive method used in the month of the interview. Women who reported OC use in the last month identified their brand of pill from a chart of OCs available in the U.S. (trademarked or registered). If the pill was not on the chart, women were asked to specify the brand or present the actual pill pack. We focused on COC use. We examined individual pill brands but also categorized COCs by: 1) ethinyl estradiol dosage (20 mcg, 25 mcg, 30 mcg, 35 mcg, or 50 mcg), 2) pill progestin component and the generation of progestin (1 st generation = norethindrone and ethynodiol acetate pills; 2nd generation = levonorgestrel and norgestrel; 3 rd generation = desogestrel, gestodene and norgestimate; 4th generation = drospirenone and dienogest), 3) monophasic or multiphasic (biphasic or triphasic) dosage; and 3) traditional 21/7 day supply versus extended or continuous cycle and shortened hormone free regimens (84/4, 24/4, 23/5, or 21/2/5 day supplies). Due to small numbers of women reporting the latter regimens, we combined them into a single category of extended/other cycle regimens. The NSFG did not assess off-label use. We used descriptive statistics to describe the sample and provide weighted estimates of the proportion of women using COCs overall and by COC type. We provide a population estimate of the overall numbers of women in the United States using COCs based upon 2010 data from the U.S. Census Bureau’s estimate of the population of women ages 15–44 years (N=62,374,964) [ 20 ]. We also estimate the proportion of women using COCs who are at risk of an unintended pregnancy (sexually active women who are not trying to become pregnant, not pregnant or not sterile [ 3 ]. We use bivariate chi-square tests to compare the proportions of women using COCs across demographic and social groups and reproductive and health characteristics. We performed multivariable logistic regression to estimate the likelihood of using types of COCs while adjusting for covariates. Covariates were considered for inclusion in regression models if their p-value in bivariate tests was ≤0.25. We present final reduced multivariate regression models in which we retained only those covariates that were significantly associated with the outcome (p<0.05). We report unweighted frequencies (n) but weighted data were used to account for the complex, stratified sampling design of the survey; weighted proportions (%), chi-square tests and odds ratios (OR) with 95% confidence intervals (CI) were computed using the svy series of commands in Stata 11.0 (Stata Corporation, College Station, TX).

Eighty-eight percent of all women aged 15–44 years (n=10,779/12,279) reported ever having used contraception; 73% had used OCs including 20% (n=2,032) in the month of the interview. Seventeen percent of all women were COC users (n=1,900); 0.4% were progestin-only pill users (n=57); 4% did not know (n=2) or report (n=73) their type of OC. Based upon the estimated 62,374,964 U.S. women ages 15–44 years, the number of current COC users is approximately 10,603,744 U.S. women. Excluding women in the survey who were pregnant (n=516), trying to become pregnant (n=395), with a history of surgical (n=1,857) or non-surgical (n=203) sterility, no sexual intercourse experience (n=1,674) or intercourse in the last 3 months (n=2,240), 25% of all women aged 15–44 at risk of unintended pregnancy were COC users (n=1,735/7,660). We describe characteristics of OC-using women (n=2,032) in Table 1 . In brief, OC-users were on average 27 ± 7 SD years old, with adolescents and young adults (15–24 years) comprising 42% of the sample and women ages 35 years and older comprising 25%. The majority of OC-using women identified as white race/ethnicity (76%), followed by Hispanic (12%), Black (9%) and other (4%) race/ethnicity. Over half the sample reported any college education (65%) and 13% were still in high school. Nearly half resided in a suburban area (48%), followed by urban (30%) and rural (21%) residences. The majority were employed (76%) but 36% reported incomes below 200% of the federal poverty level. Twenty percent of OC-using women were uninsured at some point during the previous year. Over two-thirds were currently married or cohabitating with a non-marital partner (67%). For reproductive characteristics of OC users, the average age at menarche was 12.5 years, with 8% of women reporting early age at menarche <11 years. The majority reported sexual intercourse experience (93%), with 88% reporting sex within the last three months. Less than half reported a history of pregnancy (43%); 38% had given birth to one or more children. Only 1% of OC-using women were 3 months or less postpartum and 1% were breastfeeding. History of gynecological problems was reported by 27% of OC users; these included ovulation problems (18%), ovarian cysts (14%), uterine fibroids (3%) and endometriosis (4%). For other health-related characteristics of OC users, 1% of women reported a history of non-gestational diabetes and 3% reported gestational diabetes. Nearly a quarter of OC-using women (22%) qualified as overweight with body mass indexes (BMIs) 25–29.9 kg/m 2 and 22% obese with BMIs ≥ 30 kg/m 2 . Tobacco use was reported by 18% of women; 4% smoked a pack of cigarettes per day or more. Women reported using 88 different COC brands ( Table 2 ). Among OC users (n=2,032), the five most commonly used COC brands were Ortho-Tricyclen Lo® (11%) followed by Yasmin® (11%), Yaz® (6%), and Ortho Tri-Cyclen® (5%) ( Table 2 ). Use of COCs containing ≥30 mcg of ethinyl estradiol was more common (n=1,253, 67%) than use of <30 mcg formulations (n=643, 33%); 2% (n=46) used a 50 mcg pill. Forty-one percent of OC-using women used a 3 rd generation progestin-containing COC (n=776), 22% used a 1 st generation (n=425), 19% used a 2 nd generation (n=382) and 17% used a 4 th generation (n=317) progestin pill. Norgestimate (32%, n=622) and norethindrone (20%, n=394) were the most commonly used progestins. Two-thirds of OC-using women used monophasic (n=1,255, 67%) versus multiphasic (n=645, 33%) COC dosing. Of the multiphasic COC users, the majority used a norgestimate-containing pill (n=494, 78%); half used Ortho Tri-Cyclen® or Ortho Tri-Cyclen Lo® (n=332). The majority of OC users used the traditional 21/7 day regimen pills (n=1,675, 88%); 12% used extended or other cycle regimens (n=225). In unadjusted analyses, proportions of COC use by sociodemographic and reproductive factors varied by pill type characteristic and included age, race/ethnicity, education, marital/cohabitation status, parity, postpartum status, history of gynecological problems, non-gestational diabetes diagnosis, BMI, reason for COC use and type of medical practice where the pill was received. In multivariable regression models, women aged 35–39 years (OR 1.9, CI 1.2–3.0, p=0.007) and those with a history of non-gestational diabetes (OR 11.0, CI 1.9–64.2, p=0.008) had greater odds of using a ≥30 mcg ( versus <30 mcg) pill than younger women and those without a diabetes history. Women with an underweight BMI (OR 0.3, CI 0.1–0.7, p=0.006) had reduced odds of using a ≥30 mcg dose pill compared to those with a normal weight BMI. For generation of progestin, women aged 35–39 years (OR 0.5, CI 0.3–0.8, p=0.007), with a history of gynecological problems (OR 0.7, CI 0.5–1.0, p=0.03) and an overweight BMI (OR 0.6, CI 0.4–0.8, p=0.001) had reduced odds of using 3 rd or 4 th generation progestin-containing pills ( versus 1 st or 2 nd generation pills) than women aged 15–24 years, those without gynecological problems and those with normal weight BMIs. Women with a history of gestational diabetes had greater odds of using a 3 rd or 4 th generation pill (OR 3.8, CI 1.3–10.7, p=0.01). Parous women (OR 1.7, CI 1.1–2.6, p=0.006 for parity=1 and OR 2.1, CI 1.3–3.3, p=0.002 for parity >1) and women who had received their COCs at a community (OR 1.6, CI 1.1–2.4, p=0.01) or family planning (OR 2.3, CI 1.5–3.6, p<0.001) clinic had greater odds of using multiphasic ( versus monophasic) pills than nulliparous women and women who had received their COCs at private/HMO/employer-based practices. Women who had a history of non-gestational diabetes (OR 0.1, CI 0.1–0.5, p=0.005) and those using COCs for menstrual regulation ( versus contraceptive reasons) had reduced odds of using multiphasic COCs (OR 0.6, CI 0.5–0.9, p=0.01). Finally, women with a high school diploma/GED (OR 0.3, CI 0.1–0.7, p=0.009) had reduced odds of using extended cycle regimen pills ( versus traditional 21/7 regimens) compared to those with less than a high school education. Women who used their COCs for “other” reasons ( versus for contraceptive reasons) had greater odds of using an extended/other cycle regimens (OR 2.3, CI 1.1–4.7, p=0.03).

Our finding on the prevalence of COC use among women in the U.S. is similar to national reports on contraceptive use from 1982 to 2008 [ 3 ]. Approximately 17% of our women aged 15–44 years surveyed between 2006 and 2010 were COC users (translating to a population estimate of 10,603,744 women in the United States); 25% of women at risk for an unintended pregnancy used COCs. Women used 88 different COC brands, reflecting the evolution of available modern contraceptive pills. While data were limited to type of COC use in the month of the interview (and not reflective of pill types previously tried), findings suggest that women are using earlier pill formulations (≥30 mcg estrogen, 1 st and 2 nd generation progestins, monophasic dosages and traditional 21/7 regimens) which is a departure from trends in uptake of newer formulations noted in the 1980s [ 12 , 21 ]. Women’s positive experiences with COCs they were prescribed at initiation or younger ages may provide incentive for continuation despite availability of newer formulations. Cost may also be a driving force behind this trend, given lower prices and greater insurance coverage of older COCs compared to formulations with 3 rd /4 th generation progestins and extended or other cycle regimen pills. However, these potential determinants of types of COCs used cannot be determined from these data. Reasons women did cite for current COC use were inconsistently associated with the types of pills used. Women taking COCs for acne were more likely to be using 3 rd generation pills like Ortho Tri-Cyclen® [ 2 , 22 ] but not other FDA acne-approved COCs including Estrostep® (norethindrone-based 1 st generation) or Yaz® (drospirinone-based 4 th generation). Women using COCs for “other” non-contraceptive reasons were more likely to use extended cycle regimens, which have potential therapeutic effects on endometriosis and dysmenorrhea [ 7 , 13 ], though associations were not noted between pill type and specific endometriosis and gynecological problem reasons. The NSFG does not assess off-label pill use so we likely failed to capture women who are using extended cycle regimens or other pills types for non-contraceptive purposes. Moreover, given that any COC regardless of FDA-approved indication may exhibit therapeutic effects on acne, dysmenorrhea and other conditions, the lack of association found here is not surprising. Differences in pill types across practice settings may be due to onsite availability of pharmaceutical samples, availability of inexpensive pill brands in clinic-based settings or prescription tailoring around clients’ drug prescription coverage [ 2 , 15 , 23 ]. Ortho-TriCyclen® and its equivalents (the most commonly used OCs) are widely available, commonly covered by prescription drug plans and affordable for clinic-based practices and their clientele [ 2 ]. Indeed, prescriber bias due to insurance types and formularies plays an important role in women’s use of particular pill types. We analyzed insurance status and other socioeconomic considerations potentially related to access including employment situation and income but these factors were not associated with types of COC used in multivariable models. Further evaluation of the influence of medical setting, provider prescribing patterns, and pharmaceutical access on specific COC use is warranted. While our analysis has provided some foundation as to women’s characteristics related to types of pill use, the significance of cross-sectional associations are not apparent. Limited assessment of health-related factors including cardiovascular risks did not permit adequate examination of COC use among women with hormonal contraindications, and it remains unclear whether newer pill formulations are preferable to older ones in regards to safety [ 2 , 5 , 8 , 14 , 24 , 25 – 30 ]. We were unable to assess onset, duration or lifetime use of different COC types, which is pertinent to interpreting potential relationships found here. Finally, although attempts were made to ensure accurate identification of pill type, OC use measures were self-reported and recall bias is of concern. Examination of prospective and longitudinal data including pharmaceutical, insurance claims or marketing data may help further describe types and patterns of specific OCs used over time.

Our findings suggest that women in the United States favor COCs as their contraceptive method, with the majority relying upon pills with earlier hormonal formulations. Factors determining women’s use and nonuse of lower dose, newer progestin, multiphasic or extended cycle regimen pills are likely complex and require further investigation for their role in how and why women and health providers choose to initiate or switch to particular pills. Additional prospective research is warranted to determine relationships between COC formulations and adverse events, contraceptive use patterns and unintended pregnancy rates in current contexts and among women with cardiovascular and chronic disease risk factors. Of particular interest will be the impact of socioeconomic and political factors including the forthcoming U.S. health care reform on women’s access to and use of specific types of contraception in this country. Despite the high prevalence of OC use in the U.S., long-acting reversible contraceptive methods offer an ideal option for women at risk of pregnancy when cost barriers are removed [ 31 ].