Hormonal Contraception Use and Sexual Frequency across Young Women’s Intimate Relationships

Abstract

We examined whether hormonal contraception (HC) use predicts sexual frequency throughout and across young women’s intimate relationships. From 2008–2012, the Relationships Dynamics and Social Life Study collected weekly surveys over 2.5 years, and included 893 women (aged 18–19 at baseline) who reported 2,547 intimate relationships across 32,736 weeks. Three-level logistic multilevel models assessed the weekly probability of sexual intercourse based on 1) weekly HC use (vs. non-use) and 2) duration of HC use, both accounting for several relational and individual characteristics, including relationship duration. Women had more frequent sexual intercourse when they were using HC than when they were not (predicted probabilities .65 vs .41). The weekly probability of sexual intercourse increased sharply within the first month of HC initiation (by about 27 percentage points), remained high for several months, then began to slowly decline (yet remained above that of non-use). When separated by method type, similar trajectories were found for the pill, ring, and IUD/implant; following the initial increase, steeper declines in intercourse frequency were found for the contraceptive injectable, eventually dropping below pre-initiation levels. Findings signify the immediate influence of reduced pregnancy fears in facilitating sexual intercourse among young women, which may decline as HC use continues long-term.

Sexual intimacy is a critical determinant of intimate relationship quality as well as overall health and well-being (Impett et al., 2014; Muise et al., 2016; Sprecher et al., 2018), and many young adult mixed-sex couples rely on contraceptive methods to foster sexual connection while preventing undesired or unintended¹ pregnancy. Hormonal contraception is a safe and effective method to avoid unintended pregnancy, and is the most common form of contraception among women aged 15–49 years worldwide (United Nations, 2020) and among women aged 15–29 in the U.S. (Daniels & Abma, 2020), allowing partners to enjoy sexual intercourse without the fear of unintended pregnancy. However, hormonal methods can also confer negative sexual side effects for some women (e.g., Smith et al., 2014), which may reduce sexual frequency with an intimate partner.

Sexual expression and satisfaction are important for intimacy and connection among intimate partners (Sprecher et al., 2018), and sexual problems are a top reason couples seek relational therapy (Doss et al., 2004). Given the prevalence of hormonal contraceptive use—80% of sexually experienced women in the U.S. have used the oral contraceptive pill (Guttmacher, 2020)—understanding how its use affects sexual behavior is crucial for promoting high quality intimate relationships while also promoting the prevention of unintended pregnancies. Many researchers have tried to uncover exactly how hormonal contraception may affect women’s sexuality (Burrows et al., 2012; Caruso et al., 2021; Palacios & Lilue, 2018). Yet the findings remain mixed, and we know particularly little about how young women’s use of hormonal contraception affects sexual behavior specifically in and across their intimate relationships (Higgins & Smith, 2016)—complex interpersonal contexts in which several factors are known to influence sexual behavior among partners. The aim of the current study was to document how hormonal contraception use² is linked to sexual frequency throughout young women’s intimate relationships, drawing on weekly surveys collected from a random sample of 893 women across 2.5 years.

We focus on penile-vaginal intercourse for two main reasons. First, penile-vaginal intercourse is the only sexual behavior that can result in pregnancy. Women largely use hormonal contraceptive methods so they can engage in potentially reproductive sexual behavior without the fear of an unintended pregnancy (Frost & Lindberg, 2013), and therefore, penile-vaginal intercourse is the sexual behavior most closely tied to women’s contraceptive choices. Second, penile-vaginal intercourse is strongly linked to sexual and relationship satisfaction (Sprecher et al., 2018), as well as well-being (Muise et al., 2016), among mixed-gender couples. Although penile-vaginal intercourse is but one of numerous possible sexual behaviors and only one of many sexuality-related factors that contribute to women’s sexual, relational, and personal well-being (Byers & Rehman, 2014; McClelland, 2010), it is an important indicator and outcome of women’s intimate relationships.

Young adulthood is a particularly important developmental period for women regarding sexuality and intimate relationships. For many women aged 18–22, sexual identities are continuing to form as they gain more experience with intimate and sexual relationships (Blumenstock & DeLamater, 2019; Furman, 2002; Russell et al., 2012), with these early experiences having lasting implications for later sexual and interpersonal functioning (Reissing et al., 2012; Tolman & McClelland, 2011). This also represents the age women can obtain contraceptives more autonomously, as several states (including Michigan, where the data for this analysis were collected) have restricted access to contraceptive services for minors (Guttmacher, 2021). Furthermore, women aged 18–24 report the highest rates of unintended pregnancy in the U.S. (Guttmacher, 2019) which can result in life-long detriments to upward social mobility (Kahn et al., 2014). Thus, young women represent a population who could receive maximum benefit from research that enhances our understanding of the links between hormonal contraceptives and sexual behavior with intimate partners. Below, we explain how both psychological and physiological processes may explain why hormonal contraceptive use could affect frequency of penile-vaginal intercourse among young women.

Psychological Processes

Fears of unintended pregnancy can be particularly strong deterrents to engaging in and enjoying sexual intercourse. When women in focus groups were asked about factors that influenced their sexual arousal, they described fears of unintended pregnancy as strong deterrents to arousal (Graham et al., 2004). In an interview study, a group of mixed-class young women indicated that protection from unintended pregnancy was critical for their ability to fully enjoy sexual intercourse (Higgins & Hirsch, 2008). Thus, a reduction in the fear of unintended pregnancy could increase a woman’s desire and/or willingness to engage in intercourse, or may encourage women who want to have sexual intercourse to do so (Weitzman et al., 2019), ultimately increasing a couple’s frequency of sex. Further, such reductions in pregnancy fears may also increase sexual frequency through increased enjoyment of sex.

Compared to external/male condoms—the second most commonly used contraceptive method in the United States (United Nations, 2020) and third most common among U.S. women aged 17–29 (slightly below long-acting reversible methods; Daniels & Abma, 2020)—hormonal methods are more effective in preventing pregnancy (Trussell, 2011)³. They also require less pre-planning and effort during sexual activity, and do not involve a physical barrier that can reduce sexual sensation and functioning (e.g., erectile difficulties and increased vaginal dryness) in both men and women (Higgins & Hirsch, 2008; Sanders et al., 2012; Trussell, 2009, 2011). Sexually active couples who rely on condoms to prevent pregnancy may initiate sexual intercourse more frequently following the adoption of a method that does not require effort at the time of arousal. Moreover, many young men and women believe that condoms negatively affect sexual pleasure and enjoyment and reduce feelings of intimacy during sexual intercourse (Higgins & Smith, 2016; Randolph et al., 2007) (though some also report increased feelings of safety with dual method use; Higgins & Hirsch, 2008). Couples who decide to use hormonal methods instead of condoms may have more frequent sexual intercourse due to perceptions of increased pleasure and intimacy with their romantic partner. Thus, the initiation of a hormonal method may increase sexual frequency even if the couple was using a non-hormonal method previously.

Physiological Processes

Hormones underly several physiological processes that are critical for female sexual functioning and arousability (Motta-Mena & Puts, 2017), and the extensive research on hormonal contraception use and women’s sexuality is still mixed regarding how hormonal methods may affect women’s sexual experiences (for reviews, see Burrows et al., 2012; Caruso et al., 2021; Schaffir, 2006). Some evidence indicates that hormonal contraceptive methods do not affect women’s sexuality or may even improve sexual functioning and desire among some women and couples, at least in the short-term. For example, some women have reported that oral contraceptives increased vaginal wetness and sexual pleasure (Guida et al., 2005; Veres et al., 2004), that vaginal rings improved overall sexual functioning (Caruso et al., 2014), and that intrauterine devices (IUDs) increased sexual quality of life (Skrzypulec & Drosdzol, 2008). Some have speculated that these increases could be caused at least in part by underlying psychosocial components such as decreased fear of pregnancy and the cultural importance of sexual intimacy (particularly via sexual intercourse) within romantic relationships (Palacios & Lilue, 2018).

Many studies have also found negative effects on sexuality for these methods, including oral contraceptives (e.g., reduced sexual interest; Graham et al., 1995), and vaginal rings (e.g., decreased sexual functioning; Gracia et al., 2010). Longitudinal and experimental studies have documented decreased vaginal lubrication, decreased clitoral volume, and increased pain during intercourse after initiating hormonal contraception (e.g., Battaglia et al., 2014; Burrows et al., 2012; Caruso et al., 2021). Hormonal contraception can affect hormones that influence sexual motivation such as testosterone, estradiol, and other estrogens (Burrows et al., 2012; Caruso et al., 2021; Mishell et al., 1972), all of which interact to play a crucial yet not fully understood role in female sexual desire (Cappelletti & Wallen, 2016; Caruso et al., 2021; Motta-Mena & Puts, 2017; Shirazi et al., 2019). Thus, because hormonal contraception may alter levels of hormones that have been tied to sexual desire (Caruso et al., 2021), or can make sex less enjoyable and/or more painful and thereby indirectly decrease sexual desire, young women may engage in less frequent sexual intercourse with a partner when using hormonal contraceptive methods.

A few studies have specifically documented lower sexual frequency among hormonal contraception users. In a cross-sectional study of over 1,000 women, current hormonal contraceptive users reported less frequent recent sexual activity (intercourse or otherwise) than other women (Smith et al., 2014). In their study, the statistical models included a measure of relationship sexual exclusivity (yes/no), but did not include relationship duration. Two additional studies found small decreases in sexual frequency after initiating an oral contraceptive among 22 women after 3 months (Battaglia et al., 2012) and among 48 women after 9 months (Caruso et al., 2004), but neither study included measures of the intimate relationship context. Thus, it remains unknown how sexual frequency may be linked to hormonal contraception use within relationships as they develop.

Processes over Time

The psychological and physiological consequences of hormonal contraception use may also develop over a longer period of time. While the reduction in fears of unintended pregnancy may be immediate, the learning process whereby increased sexual enjoyment leads to more frequent sex may take more time, as it often takes multiple experiences to form strong and reliable associations that can result in sexuality changes (Hoffmann, 2017). Similarly, negative sexual side effects that could decrease sexual frequency may not occur immediately following hormonal contraception initiation. In the small study of 48 women initiating a low-dose oral contraceptive, sexual desire and frequency of intercourse did not decline until 9 months after baseline, but sexual enjoyment decreased after as little as 3 months (Caruso et al., 2004). Though it is unclear whether these findings would hold after accounting for relationship duration, this suggests a possible delay in reduced sexual enjoyment translating to less frequent sexual behavior. If couples also acclimate over time to their reduced contraceptive effort at the time of intercourse, we may see a short-term increase (due to reduced fear of pregnancy) but a long-term decrease in sexual frequency after hormonal contraception is initiated.

The Current Study

Our first research aim was to assess whether young women were more likely to engage in sexual intercourse when they were using versus not using hormonal contraception. Our second research aim was to assess whether frequency of sexual intercourse changed over the course of hormonal contraception use. Our overarching hypothesis was based on both psychological and physiological mechanisms. We expected that the short-term effect of adopting a hormonal contraceptive method would be psychological—couples’ immediate protection against unintended pregnancy, or immediate reduction in fear of unintended pregnancy, would result in an immediate increase in the weekly probability of sexual intercourse. However, we also predicted that couples would acclimate to these psychological changes over time, and that physiological mechanisms may become more important as hormonal contraceptive use continues. Thus, we expected to see a long-term decrease in sexual frequency with continued hormonal contraceptive use.

Because of the unknown specific hormonal configurations of the hormonal methods used and because some methods were used quite infrequently, our primary analyses focused on hormonal contraception methods broadly (i.e., any hormonal method), similar to previous studies (Blumenstock & Papp, 2020; Smith et al., 2014). However, because different methods may confer different effects, we also explored the hypothesized associations separately by method type.

The Relationship Dynamics and Social Life (RDSL) dataset (Barber et al., 2016) offers a unique opportunity to enhance our understanding of these links between hormonal contraception and sexual frequency. First, RDSL collected weekly surveys for 2.5 years, resulting in multiple assessments within each intimate relationship that lasted more than one week, and for most women, multiple intimate relationships. The repeated surveys within each woman’s relationship(s) allow us to disaggregate the overall associations between hormonal method use and sexual frequency into within-relationship differences (i.e., sexual frequency during weeks in which a woman uses hormonal contraception versus weeks she does not), between-relationship differences (i.e., sexual frequency across a woman’s relationships that include more versus less hormonal contraception use), and between-woman differences (i.e., sexual frequency of women who use hormonal contraception more regularly or for longer periods of time versus women who use it less). Separating such within versus between effects also disentangles selection effects. For instance, mixed-sex couples with the highest sexual frequency may also be most likely to choose hormonal contraception precisely because they recognize that their risk of unintended pregnancy is higher than couples who have less frequent sexual intercourse. Thus, disentangling such selection effects is critical for fully understanding how hormonal contraception use corresponds to couples’ sexual frequency.

Second, much of the research examining the impact of hormonal contraception on sexual outcomes has disregarded the intimate relationship context. Some analyses have focused only on women in specific types of relationships (e.g., cohabiting relationships; Caruso et al., 2014), which avoids confounding by relationship type, but limits the generalizability of results. In addition, most of these studies have not statistically accounted for relationship duration, despite the connection between longer relationship duration and lower sexual frequency (James, 1981; Jasso, 1985; Schröder & Schmiedeberg, 2015; Sprecher et al., 2006). In one exception, a small diary study of 43 couples indicated hormonal contraception use (broadly) was associated lower sexual enjoyment for women and their male partners, and the analyses accounted for relationship satisfaction and duration (Blumenstock & Papp, 2020). The RDSL’s intensive longitudinal design included all intimate relationships, regardless of type, and measured changes in those relationships over time (e.g., cohabitation, engagement, marriage), as well as duration. This allows us to disentangle changes in sexual frequency that occur as a result of initiating or continuing hormonal contraception from changes that may be due to couples transitioning to new relationship stages (e.g., moving in together) or to declines in sexual frequency as a relationship continues (James, 1981; Jasso, 1985; Schröder & Schmiedeberg, 2015; Sprecher et al., 2006).

Third, other than small, short-lived (usually 9 months or shorter) clinical studies, very little research has examined young women’s sexual behaviors over the entire course of a contraceptive episode (initiation to discontinuation), or when use has been longer than a clinical study period. Further, clinical studies may not represent young women’s naturalistic use of contraceptive methods, such as their decisions to initiate or discontinue use throughout their daily lives and their intimate relationships. The current study captures young women’s naturalistic contraceptive episodes throughout their intimate relationships and examines how duration of hormonal contraceptive use is linked to their frequency of sexual intercourse with a partner.

Method

Participants and Procedure

Data were drawn from the Relationship Dynamics and Social Life (RDSL) study, an intensive longitudinal study designed to assess young women’s intimate relationships and sexual and pregnancy experiences (Barber et al., 2016). The population-based sample included women aged 18–19 who were randomly selected from Genesee County in Michigan, a racially and socioeconomically diverse community. Women were sampled via driver’s license and Personal Identification Card databases. When compared with census population projections, this sampling frame indicated a 96% agreement (Barber et al., 2011). The response rate was 83%.

The RDSL study involved two main parts, an initial baseline interview and weekly follow-up surveys. The initial baseline interviews, conducted from 2008–2009, involved face-to-face interviews with 1,003 women. This baseline interview lasted approximately one hour and included assessments for sociodemographic characteristics, intimate relationship experiences, and attitudes and experiences related to sexuality, contraception, and pregnancy. The weekly follow-up surveys spanned 2.5 years following the baseline interview and assessed participants’ intimate relationships, contraception use, sexual intercourse, and desires for and experiences with pregnancy. The weekly surveys were administered online, or sometimes via phone when the respondent did not have internet access. Almost all women (992; 99%) participated in the follow-up study phase; 84% remained in the study for at least 6 months and 75% remained for at least 18 months. In all, 58,594 weekly surveys were completed. Compensation included $5 sent in advance and $30 for completing the baseline interview. Additional incentives were provided for participating in the weekly surveys, including $5 per survey for the first four weeks and $1 per survey afterwards, with $5 bonuses for on-time completion of five interviews in a row.

Of the 992 women who participated in the weekly surveys, 925 reported at least one intimate relationship. Weeks during which women reported being pregnant or a high desire for pregnancy were removed from our analysis, resulting in a final sample of 32,736 weeks. After removing the pregnancy weeks, sexual intercourse data were not available for 32 women. Thus, the current sample included the 893 (90.0%) women who reported an intimate relationship and provided data on sexual frequency.

Among the final sample of 893 women, the average age was 19.2 years (SD = 0.57, range 18–20). Most (511, 57.2%) identified as White, 265 (29.7%) as African American, 108 (12.1%) as multiple races, and less than 1% as either American Indian, Asian, or Pacific Islander, or did not answer. Other individual and relationship characteristics are presented in Table 1.

Table 1.

Descriptive Statistics for Woman and Relationship Characteristics

Variable	N	%	M	SD	Range
Woman characteristics
N (women)	893
Intercourse prior to age 17	234	26.2
Two or more sexual partners	548	61.4
Previous pregnancies			0.30	0.60	0–2
Childhood disadvantage index			1.29	1.12	0–4
Mother first birth <19	322	36.1
Mother <high school education	77	8.6
Non-two-parent household	425	47.6
Public assistance	328	36.7
African American	309	34.6
Highly religious	514	57.6
Receiving public assistance	232	26.0
High school GPA			3.11	0.61	0–4.17
Pregnancy avoidance			4.78	0.54	1–5
Frequency of condom usea			.42	.39	.00–1.00
Number of weekly surveys			36.5	34.1	1–146
Number of relationships			2.92	2.8	1–23
Relationship characteristics
N (relationships)	2,547
Relationship length (months)			9.78	16.89	0.08–131.9
Participant age (at relationship start)			19.89	0.94
Age differenceb			2.22	3.56	−5.9 – 33.2
Couple had a shared birth	194	7.6
Couple had shared pregnancy	362	14.2
Partner characteristics
Education (years)			12.50	1.09	10–14
African American	917	36.0
Relationship type (# weeks)
Casual	2,354	7.2
Uncommitted dating	3,153	9.6
Committed dating	16,270	49.7
Cohabiting	5,257	16.1
Engaged	4,092	12.5
Married	1,610	4.9
Number of weekly surveys			12.77	22.89	1–146

^aProportion of condom-use weeks by total intercourse weeks.

^bPartner age minus participant age, in years.

Measures

In each weekly survey, women were asked a series of questions to determine whether they had any type of intimate partner in the prior week (e.g., a romantic partner, any physical or emotional contact, including anyone with whom they had sexual intercourse). In the 1% of weeks when women had more than one partner, they referred to their primary or most serious partner in the survey questions. Specific partners were tracked over time—including those that broke up and later reconciled—using initials or a nickname provided by the women.

Sexual frequency.

Each week, women were asked whether they had penile-vaginal intercourse during the past week (coded 1=yes, 0=no): “In the past {days since last survey} since {date of last survey}, did you have sexual intercourse with {current partner’s name}? By sexual intercourse, we mean when a man puts his penis into a woman’s vagina.” They were not asked the number of times they had sex during the week. Thus, the RDSL includes a specific measure of sexual frequency—a time-varying indicator of whether the couple had sex at least once each week. For a relationship as a whole, the proportion of weeks that included sexual intercourse indicates how regularly the couple had sexual intercourse throughout that relationship.

Having sexual intercourse weekly represents a meaningful frequency in at least two ways. First, sexual frequency is linearly associated with higher well-being for frequencies up to once per week, but there are no differences in well-being at frequencies greater than once per week (Muise et al., 2016). Second, based on the General Social Survey, the median sexual frequency for women ages 18–24 is weekly (Ueda et al., 2020).

Hormonal contraception use.

Each week, women were asked “Did you use anything that can help people avoid becoming pregnant, even if you did not use it to keep from getting pregnant yourself?” If yes, they were asked whether they used any of the following specific types of contraception: birth control pills, birth control patch, the NuvaRing, Depo-Provera or any other type of contraceptive shot, an implant such as Implanon™ or another contraceptive implant, or an IUD. For the main analyses, these were coded as 1 = any hormonal method use, 0 = no hormonal method use. For the exploratory subanalyses by method type, variables were coded similarly (e.g., pill use = 1, no pill use = 0). Responses for IUD and Implant were combined to form a category for long-acting reversible contraception (LARCs). (Although some IUDs do not contain hormones, RDSL did not distinguish between IUD types; only 12 women used an IUD at any point during the study period.)

Duration of hormonal contraception use.

A time-varying measure of the cumulative number of months the woman used the hormonal contraception method was computed as the total number of days she reported using the specific method, divided by the average number of days in a month (30.4). For the first week of method use, we assumed the method was started at the mid-point since her last report (e.g., if her previous report was 7 days prior, total days of use that week was calculated to be 3.5 days). For subsequent weeks (in which the method was reported the prior week and the current week), we assumed the method had been used for all days since the prior survey. When a weekly survey indicated she was no longer using the method, we assumed the method was discontinued at the mid-point between the prior survey and current survey. If a method was discontinued and re-started, duration re-started at 0. The time-varying cumulative duration variable was then recoded into 8 separate indicator variables to fully capture any non-linear effects: No use (or prior to initiation), <1 month, 1 month to <3 months, 3 months to <6 months, 6 months to <9 months, 9 months to <12 months, 12 months to <18 months, 18 months or greater. Each week was coded into one of these categories.

For the exploratory subanalyses by method type, we coded a similar set of time-varying indicator variables. Because of the infrequent use (and thus small number of weeks) of some methods, we condensed the six duration indicator variables to three to increase the frequencies in each category while maintaining flexibility and precision: No use (or prior to initiation), <1 month, 1 month to <6 months, 6 months to <12 months, 12 months or greater. These same five indicators were used for each method type for comparability.

Relationship characteristics.

Multiple characteristics of relationships that are associated with sexual activity were included in the models, including weekly-varying indicators of commitment levels, which distinguished between relationship partners that were married, engaged to be married, sharing a residence (cohabiting), in a serious dating relationship but not cohabiting, or in a casual relationship. Weekly time-varying relationship duration was calculated each week, in months, by summing the number of days with the partner and dividing by the average number of days in a month (30.4). If a woman was in a relationship at study onset, she was asked the month and year it began (we assumed the relationship began at the mid-point of the month). If a new partner was reported in a weekly survey, we assumed it began at the mid-point between the current and previous weekly surveys. Relationship duration was included for all participants, whether or not they were using a hormonal contraceptive. Partner demographic characteristics included African American/Black racial identity (yes = 1), age difference in years, and education in years. Relationship-level shared childbearing was coded as yes=1, no=0.

Woman characteristics.

The models also included relevant individual-level background variables, assessed at the baseline survey interview. Adolescent sexual/reproductive history included early first intercourse (coded as <17 = 1, otherwise 0), number of sexual partners (coded as 2+ = 1, <2 = 0), and number of prior pregnancies (as reported; range 0–2). Family background and childhood disadvantage were measured as a sum of four indicators, including mother’s first birth at a young age (coded as <19 years old = 1), mother’s low educational attainment (coded as less than high school = 1), grew up in non-two-parent household (coded as yes = 1), and received public assistance during childhood (coded as yes = 1). Demographic characteristics included self-reported African American/Black racial identity (coded yes = 1), high religiosity (coded as religious faith being very important or more important than anything else = 1, somewhat or not important = 0), currently receiving public assistance at baseline (coded as yes = 1), and high school grade point average (GPA; coded as reported value).

Condom use was also included. In weekly surveys, if a woman indicated sexual intercourse occurred that week, she was asked whether a condom was used. Because condom use was only reported if sexual intercourse occurred, this could not be included as a week-level predictor (i.e., there was no variability because every week of condom use also included sexual intercourse). Frequency of condom use was calculated as the proportion of the total weeks of sexual activity that included condom use.

Data analysis

The repeated measures data was nested in two ways, resulting in three levels: weeks nested within relationships, and relationships nested within women, with weeks as level 1, relationships as level 2, and women as level 3. Multilevel modeling can simultaneously estimate effects across weeks, across relationships, and across women, while accounting for the inherent interdependence within each level (Raudenbush & Bryk, 2002). Three-level generalized multilevel models (GMLM) with Bernoulli distribution and logit link function were used to assess the probability of sexual intercourse as a function of contraception use.

For our first research aim (differences in sexual frequency by weekly hormonal contraception use vs. non-use), week-level hormonal contraceptive use was entered as a time-varying Level 1 predictor. For our second research aim (changes in sexual frequency across the duration of hormonal contraceptive use), time-varying indicators of duration of hormonal contraceptive use were entered as Level 1 predictors of sexual intercourse; the reference category was duration <1 month (i.e., immediately after initiation). For both models, Level 1 (week level) covariates included relationship duration and commitment levels, Level 2 (relationship level) covariates included partner demographics and relationship-level shared childbearing, and Level 3 (woman level) controls included adolescent sexual/reproductive history, family background and childhood characteristics, and demographic variables. Time-varying length of relationship was also included in all models to assess the effects of hormonal contraception use and duration net of the effect of relationship duration on the probability of sexual intercourse.

To disentangle selection effects (i.e., to separate within-subjects effects from between-subjects effects; Hoffman, 2015), between-relationships (level 2) and between-women (level 3) variables for hormonal contraception use were calculated and included for both models. For the use vs. non-use model, the between-relationship variable was calculated as the overall proportion of weeks in the relationship that included hormonal contraceptive use, and the between-woman variable was the overall proportion of the woman’s total weeks that included hormonal contraception use. For the duration models, the between-relationships variable was calculated as the average total duration of each hormonal contraception episode within the relationship, and the between-woman variable was calculated as the woman’s average total duration of each hormonal contraception episode throughout the study period. The between-relationships variables were group-mean-centered, and therefore the coefficients indicate differences in relationship-level sexual frequency as a function of whether the relationship was above or below a woman’s average hormonal contraception episode across all her relationships. Between-women variables were grand-mean-centered, and therefore the coefficients indicate woman-level differences in sexual frequency associated with her hormonal contraceptive use compared to the sample average hormonal contraceptive use (i.e., higher or lower than the average across all women).

To investigate potential nuances in the results, two subanalyses were conducted. First, because women who used hormonal contraceptive methods for longer durations may differ from those who discontinued use earlier, the duration models were also run with only the women who had used a method for 12 months or longer (n = 150). Second, as mentioned above, exploratory analyses were also conducted, with both models estimated separately for each method type.

Results

Descriptive information for sexual intercourse and hormonal contraception is presented in Table 2. Women used hormonal contraception in 43.1% of weeks (women’s percentages ranged from 0–100%). In terms of duration, women used hormonal contraception during the study period for anywhere from 0.1 (just starting) to 31.4 months (M = 7.7, SD = 8.7). Most (87.1%) hormonal contraceptive episodes occurred within a single relationship (i.e., were initiated and discontinued within a single relationship), and most relationships (70.3%) included only one hormonal contraceptive episode (i.e., women initiated, discontinued, and re-initiated a hormonal method in less than one-third of relationships). Descriptive information of hormonal contraception use by method type is presented in Table 3. The pill was used by over the half the women (57.0%) and was by far the most frequently reported method, followed by the injectable (15.1%), the IUD or implant (7.1%), the ring (5.4%), and the patch (4.2%).

Table 2.

Descriptive Statistics for Sexual Intercourse and Hormonal Contraception Use During Women’s Intimate Relationships

	Women		Relationships		Weeks
	M(SD) or proportion	Range	M(SD) or proportion	Range	Total weeks/proportion
Total N	893		2,547		32,736
Sexual intercourse
Any	0.83		0.63		0.54
Total weeks	19.1(23.6)	0–124	6.69(15.2)	0–124	17,042
Proportion of total weeks	0.52(0.34)	0–1	0.45(0.42)	0–1
Hormonal contraception
Any	0.68		0.46		0.43
Total weeks	15.6(25.7)	0–146	5.35(15.3)	0–146	13,725
Proportion of total weeks	0.37(0.38)	0–1	0.33(0.42)	0–1
Number of episodes	1.92(2.4)	0–19	1.65(1.5)	0–16
Total duration in monthsa	3.7(6.4)	0.1–31.8	6.3(7.8)	0.1–31.4
Duration categories
No use					0.57
Less than 1 month					0.09
1–3 months					0.09
3–6 months					0.08
6–9 months					0.05
9–12 months					0.04
12–18 months					0.04
More than 18 months					0.04

^aAverage of episodes across all women’s episodes, or all episodes within the relationship

Table 3.

Descriptive Statistics for Hormonal Contraception Use by Method Type

	Pill	Patch	Ring	Injectable	IUD or Implant
Women reporting any use, N(%)	520(57.0%)	39(4.2%)	50(5.4%)	139(15.1%)	65(7.1%)
Relationships with any use, N(%)	977(37.1%)	40(1.5%)	69(2.6%)	190(7.2%)	107(4.1%)
Weeks reporting method
M(SD)	21.5(28.0)	8.6(18.1)	9.7(20.5)	9.8(16.3)	10.6(12.9)
Range	1–146	1–82	1–102	1–122	1–60
Total duration in monthsa
M(SD)	4.19(7.0)	3.29(6.2)	2.05(4.2)	2.65(5.1)	3.00(4.8)
Range	0.1–31.0	0.1–27.7	0.1–25.9	0.1–31.8	0.1–22.4
Total weeks, N	10,851	328	475	1,310	659
Duration categories, N
Less than 1 month	2000	67	129	361	201
1–3 months	2250	66	95	317	147
3–6 months	1980	65	55	271	125
6–9 months	1254	49	56	123	85
9–12 months	926	31	55	80	51
12–18 months	1164	34	55	114	32
More than 18 months	1277	16	30	44	18

^aAveraged across all episodes

Hormonal Contraception Use vs. Non-Use

Use vs. non-use model results for hormonal contraception use broadly (i.e., any hormonal method) from the logistic multilevel models are presented in Table 4 (full model results available in supplemental materials). Week-level (within-relationships; level 1) results indicated that sex was more probable while women were using a hormonal contraceptive than while they were not; women had over three times higher odds (AOR=3.41, p < .001) of engaging in sexual intercourse during the weeks they used a hormonal method compared to the weeks they did not (predicted probability of .65 vs .41, holding all other variables at their means). Relationship-level (between-relationships, level 2) results indicated no significant difference in the probability of sex in a woman’s relationships in which hormonal contraceptive use was more versus less frequent, whereas between-woman (level 3) results indicated that women who more frequently used hormonal contraception (i.e., a higher proportion of weeks with hormonal contraceptive use) had higher weekly odds of sex (AOR=1.36, p < .001). In other words, women who used hormonal contraception more regularly also had more regular sexual intercourse, but sexual frequency within a woman’s relationship did not differ as a function of the overall level of hormonal contraception use within that relationship.

Table 4.

Hormonal Contraceptive Use and Duration Predicting Weekly Sexual Intercourse: Results from 3-Level Generalized Multilevel Models

	Any use vs. No use					Duration of use
	B	SE	p	AOR	AOR 95% C.I.	B	SE	p	AOR	AOR 95% C.I.
Intercept	−2.15	0.11	<.001	−2.15	(0.093,0.145)	−0.91	0.12	<.001	0.40	(0.314,0.513)
Woman level (Level 3)
Hormonal Contraception
Proportion of weeks used	0.31	0.09	<.001	1.36	(1.149,1.611)
Avg. total duration of usea						0.00	0.01	.606	1.00	(0.987,1.008)
Relationship Level (Level 2)
Hormonal Contraception
Proportion of weeks used	−0.20	0.15	.191	0.82	(0.610,1.104)
Avg. total duration of usea						0.03	0.01	.002	1.03	(1.011,1.050)
Week level (Level 1)
Hormonal contraception
Used that week	1.23	0.06	<.001	3.41	(3.049,3.805)
Duration of useb
No use						−1.23	0.07	<0.001	0.29	(0.255,0.332)
<1 month (Reference)						-	-	-	1.00	-
1–3 months						0.05	0.08	0.535	1.05	(0.896,1.235)
3–6 months						−0.10	0.09	0.244	0.90	(0.762,1.072)
6–9 months						0.08	0.10	0.449	1.08	(0.884,1.322)
9–12 months						−0.03	0.11	0.769	0.97	(0.775,1.207)
12–18 months						−0.15	0.11	0.190	0.86	(0.694,1.075)
18+ months						−0.37	0.12	0.003	0.69	(0.543,0.887)

Note. Avg. = Average. B = Unstandardized coefficients. SE = Standard error. AOR = Adjusted odds ratio. Full results available in supplemental materials.

^aMonths; averaged across each episode within woman (Level 3) or within relationship (Level 2).

^bTime-varying indicator variables.

Several relationship variables were also associated with sexual intercourse in the model. Specifically, greater age difference between partners, partner of African American race, and any pregnancy during the relationship were all associated with more frequent intercourse during the relationship, and frequency of intercourse declined as relationship length increased (all p values < .003; full results in supplemental materials). In general, the model demonstrated higher probability of intercourse when commitment was stronger (as indicated by relationship status; all p values < .001; full results in supplemental materials).

Subanalyses of specific contraceptive method types for the use vs. non-use models also indicated the same patterns—the probability of sexual intercourse was significantly higher when women were using the hormonal method versus when they were not using the method (key results presented in Table 5; full results available in supplemental materials). The only non-significant (positive) association was for the contraceptive patch, which was the least frequently reported method, suggesting low statistical power for finding small effects.

Table 5.

Hormonal Contraceptive Use Vs. Non-Use Predicting Sexual Intercourse, by Method Type: Results from 3-Level Generalized Multilevel Models (5 Separate Models for the 5 Method Types)

	B	SE	p	AOR	AOR 95% C.I.
Woman level (Level 3)
Proportion of weeks used
Pill	0.25	0.09	0.006	1.28	(1.071,1.526)
Patch	0.44	0.77	0.573	1.55	(0.339,7.052)
Ring	0.38	0.26	0.136	1.47	(0.886,2.436)
Injectable	0.34	0.19	0.083	1.40	(0.956,2.054)
IUD/Implant	−0.23	0.37	0.535	0.80	(0.389,1.633)
Relationship Level (Level 2)
Proportion of weeks used
Pill	−0.01	0.15	0.974	0.99	(0.735,1.347)
Patch	−1.35	0.86	0.118	0.26	(0.048,1.410)
Ring	0.67	0.55	0.226	1.95	(0.661,5.773)
Injectable	−0.08	0.48	0.868	0.92	(0.361,2.362)
IUD/Implant	−0.81	0.55	0.142	0.44	(0.151,1.314)
Week level (Level 1)
Used that week (yes/no)
Pill	1.04	0.06	<0.001	2.83	(2.513,3.180)
Patch	0.50	0.27	0.067	1.65	(0.965,2.821)
Ring	0.74	0.19	<0.001	2.10	(1.445,3.065)
Injectable	0.57	0.12	<0.001	1.77	(1.404,2.221)
IUD/Implant	1.56	0.21	<0.001	4.76	(3.128,7.250)

Note. B = Unstandardized coefficients. SE = Standard error. AOR = Adjusted odds ratio. Five separate models were run, one each for the five different method types, with the upper-level variables corresponding to that method type. Bold font indicates significant level-1 associations. Full results available in supplemental materials.

Hormonal Contraception Duration

Duration model results for hormonal contraception use broadly are also presented in Table 4 (full model results available in supplemental materials). Figure 1 presents the predicted probabilities by duration based on these models, illustrating how the probability of sexual intercourse changed over time as women initiated and continued to use hormonal contraception in their relationships. Figure 1 illustrates that the probability of sex increased sharply, from 43% to 70%, during the first month after hormonal contraceptive initiation, remained high for several months, and then began to decrease slowly. The probability never returned to the pre-initiation level, however. Relationship-level (between-relationships, level 2) results indicated that relationships with longer hormonal contraceptive use, relative to average use across all of a woman’s relationships, also had more frequent sexual intercourse (AOR=1.03, p = .002) compared to relationships with below-average hormonal contraception use. Between-women (level 3) results indicated no significant association between a woman’s average length of hormonal contraception use and her frequency of sexual intercourse (p = .606). In contrast to the use vs non-use model, the upper-level connections between sexual frequency and duration of hormonal contraception use were stronger at the relationship level than at the woman level.

Figure 1.

Predicted Probabilities of Weekly Sexual Intercourse over the Course of Hormonal Contraception Use, with Main Analysis and Subanalysis of Women who Used Hormonal Contraception for 12 Months or Longer

Note. HC = Hormonal contraception. Results from 3-level multilevel models accounting for several relationship characteristics, including duration. Values closer to 1 represent a higher probability of sexual intercourse. Sample with all HC episodes (i.e., the full sample) included N = 893 women; subsample of women who used hormonal contraception for 12 months or longer included n = 150 women.

Several relationship variables were also associated with sexual intercourse, which were consistent with the use vs. non-use models. Specifically, greater age difference between partners, partner African American race, and pregnancy in the relationship were all associated with more frequent intercourse in the relationship, and increases in relationship length were associated with decreases in the probability of sexual intercourse (ps < .004; full results in supplemental materials). Relationship status indicators generally indicated greater probability of intercourse with increases in commitment levels (ps < .001; full results in supplemental materials).

Subanalyses including only those who used a hormonal contraception method for longer than 12 months (n = 150) indicated the same patterns (also plotted in Figure 1). Results presented in Table 6; full model results available in supplemental materials.

Table 6.

Subanalysis Including Only Women Who Had Used Hormonal Contraception for 12 Months or Longer: Duration Variable Results from 3-Level Generalized Multilevel Models

Duration variable (time-varying)	B	SE	p	AOR	AOR 95% C.I.
No use	−1.40	0.15	<0.001	0.25	(0.183,0.330)
<1 month (reference)	-	-	-	1.00	-
1 to <3 months	−0.01	0.14	0.961	0.99	(0.757,1.303)
3 to <6 months	−0.21	0.14	0.130	0.81	(0.621,1.063)
6 to <9 months	−0.03	0.15	0.818	0.97	(0.727,1.286)
9 to <12 months	−0.12	0.15	0.441	0.89	(0.665,1.195)
12 to <18 months	−0.15	0.15	0.291	0.86	(0.643,1.142)
18+ months	−0.36	0.16	0.027	0.70	(0.507,0.959)

Note. B = Unstandardized coefficients. C.I. = Confidence interval. SE = Standard error. AOR = Adjusted odds ratio. Full results available in supplemental materials. N = 8,339 weeks among 150 women.

The exploratory subanalysis models by method type also indicated a marked increase in the probability of sexual intercourse during the first month the method was initiated, and this largely remained elevated across duration for the contraceptive pill, ring, and IUD/implant (results presented in Table 7; plotted in Figure 2). For these three methods, the multilevel models indicated the immediate increase was statistically significant at p < .05. However, any changes after this initial increase were not statistically significant (all ps ≥ .417), suggesting the probability of sexual intercourse remained similarly high once elevated, despite descriptive decreases for the contraceptive ring and increases for the pill and IUD/implant. For the contraceptive injectable, results indicated a similarly steep and significant increase immediately following initiation, yet the probability of intercourse declined more steeply, quickly returning to pre-initiation levels and below (probabilities at all subsequent time points were significantly lower than in the first month). Lastly, the contraceptive patch was reported too infrequently for the duration model to be estimated.

Table 7.

Hormonal Contraceptive Use Duration Predicting Sexual Intercourse: Duration Variable Results from Separate Models by Method Type using 3-Level Generalized Multilevel Models

Duration variable (time-varying)	B	SE	p	AOR	AOR 95% C.I.
Pill only
No use (prior to initiation)	−0.62	0.10	<.001	0.54	(0.443,0.647)
<1 month (reference)	-	-	-	1.00	-
1 to <6 months	0.04	0.08	.614	1.04	(0.887,1.226)
6 to <12 months	0.10	0.13	.417	1.11	(0.864,1.422)
12+ months	−0.03	0.16	.840	0.97	(0.708,1.324)
Ring only
No use (prior to initiation)	−0.82	0.39	.037	0.44	(0.205,0.952)
<1 month (reference)	-	-	-	1.00	-
1 to <6 months	−0.17	0.38	.664	0.85	(0.402,1.786)
6 to <12 months	−0.36	0.58	.533	0.70	(0.223,2.171)
12+ months	−0.60	0.76	.425	0.55	(0.124,2.409)
Injectable only
No use (prior to initiation)	−0.71	0.14	<.001	0.49	(0.371,0.652)
<1 month (reference)	-	-	-	1.00	-
1 to <6 months	−0.45	0.15	.003	0.64	(0.474,0.858)
6 to <12 months	−0.53	0.24	.026	0.59	(0.369,0.939)
12+ months	−1.22	0.35	<.001	0.29	(0.148,0.585)
IUD/Implant only
No use (prior to initiation)	−0.94	0.40	.017	0.39	(0.179,0.846)
<1 month (reference)	-	-	-	1.00	-
1 to <6 months	0.15	0.21	.465	1.17	(0.771,1.766)
6 to <12 months	0.26	0.28	.348	1.30	(0.753,2.236)
12+ months	0.23	0.27	.386	1.26	(0.746,2.132)

Note. B = Unstandardized coefficients. C.I. = Confidence interval. SE = Standard error. AOR = Adjusted odds ratio. Full results available in supplemental materials. Due to low frequencies of reported use for some method types, duration indicator variables were condensed to 5 categories from the original 8 to ensure sufficient data for valid analyses (at least 100 reports in most categories). Even with the condensed categories, the contraceptive patch was reported too infrequently for this model to be estimated.

Figure 2.

Predicted Probabilities of Weekly Sexual Intercourse over the Course of Hormonal Contraception Use: Subanalyses by Method Type, with Combined Model Included for Reference

Note. Results from 3-level multilevel models accounting for several relationship characteristics, including duration. Values closer to 1 represent a higher probability of sexual intercourse. Due to relatively infrequent reporting of several methods, latter duration categories were combined to ensure sufficient data in each; patch is not included because reporting was too infrequent to estimate the model. The model with all methods combined (i.e., the primary analysis, also presented in Figure 1) is presented here as a reference for easier comparison.

Discussion

Our findings were largely consistent with our hypothesis that the initiation of hormonal contraception would increase the probability of sexual intercourse, but that the effect would dissipate over time. Indeed, women engaged in sexual intercourse more frequently when they were using hormonal contraceptives than when they were not, and this was true whether use was assessed broadly or separately by method type. The probability of sexual intercourse increased sharply during the first month after hormonal contraception was initiated. When assessing hormonal contraceptive use broadly (i.e., any method), the probability remained elevated compared to pre-initiation levels, but it decreased slowly, above and beyond decreases in the probability of sexual intercourse as relationships endure (Jasso, 1985; Schröder & Schmiedeberg, 2015; Sprecher et al., 2006). When separated by method type, a similar pattern emerged for three of the methods (pill, ring, IUD/implant), with an immediate increase that remained above pre-initiation levels. However, the contraceptive injectable indicated a diverging pattern, with the immediate increase in sexual frequency followed by quicker and steeper declines.

For the contraceptive pill, ring, and IUD/implant, such immediate and prolonged increases in the frequency of sexual intercourse over the course of hormonal contraception use support the key proposed psychological mechanism—reduced fears of unintended pregnancy—underlying the links between hormonal contraception use and sexual behavior (Graham et al., 2004; Higgins & Hirsch, 2008). This is consistent with women’s descriptions of their sexual and contraceptive experiences, which underscore reliable pregnancy prevention as a requirement for engaging in and enjoying sexual activity with a partner (Graham et al., 2004; Higgins & Hirsch, 2008). Yet, for the combined models, after about 12 months of continuous hormonal contraceptive use—a time frame that extends beyond what many longitudinal studies have been able to capture—sexual frequency began to decline, which is consistent with the physiological mechanisms we proposed. When assessed separately, sexual frequency when using the contraceptive pill, ring, and IUD/implant remained elevated, with some visual and descriptive evidence of declines that were not statistically significant. This could indicate a lack of negative sexual side effects during the first year of hormonal contraception use, or a delayed effect. The psychological effects (e.g., mainly reduced fear of unintended pregnancy, but also potentially perceptions of closeness, resultant increased sexual enjoyment, cultural emphasis on sexual intercourse for sexual intimacy) may have outweighed any sexual side effects at first, then subsided as couples acclimated to the effects. These results also echo recent findings from a longitudinal 3-month study indicating that most women experienced improvements or no changes to their sex lives following initiation of a hormonal contraceptive method or a copper IUD (Higgins et al., 2021).

Alternatively, women may choose to engage in intercourse with a romantic partner despite any negative sexual side effects of a contraceptive method. Women are more likely than men to minimize their own experiences of pleasure or discomfort in service of maintaining sexual intimacy with a (male) partner (Mark et al., 2014; McClelland, 2010), which could also partially explain such robust and sustained increases in sexual frequency despite a large body of evidence suggesting hormone-related declines in sexual functioning for some (though not all) women (Burrows et al., 2012; Caruso et al., 2021; Schaffir, 2006.

Conversely, when women were using the contraceptive injectable, which is a progesterone-only method, the immediate increase in sexual intercourse frequency was quickly followed by a subsequent decline. This could indicate potential negative sexual side effects, as the probability of intercourse declined sharply, eventually dropping below pre-initiation levels. This is consistent with some studies indicating multiple adverse effects of the contraceptive injectable on women’s sexual experiences, such as weight gain, decreased libido, and heightened emotionality (reviewed in Higgins & Smith, 2016). At 6-month follow-ups among 1,900 women initiating a contraceptive method, a lack of sexual interest was more common among those using the Depo Provera injectable compared to other hormonal methods (Boozalis et al., 2016). The contraceptive injectable is a progesterone-only method, and these findings offer limited support to other preliminary evidence that progesterone may inhibit desire (while estrogen may promote desire; Roney & Simmons, 2013). However, the uncertainty in the dataset about the specific hormonal content of each method impedes drawing any conclusions about differences based on hormone content, and encourages future research via future large, naturalistic studies of women’s sexuality and hormonal contraception use to assess more detailed information about each method’s hormonal composition.

Our results contrast with cross-sectional differences in sexual frequency, in which women who were using hormonal contraceptive methods reported lower recent sexual frequency than those not using hormonal methods (Smith et al., 2014). Particularly considering the physiological side-effects of hormonal contraceptive use reported among many women—decreased vaginal lubrication, decreased clitoral volume, and increased pain during heterosexual intercourse (Battaglia et al., 2014; Burrows et al., 2012; Caruso et al., 2021)—these cross-sectional differences have reasonably been interpreted dynamically, as evidence that sexual frequency declines with the adoption of a hormonal contraceptive method. However, our analysis of dynamic data shows that a couple’s sexual frequency is actually higher when using a hormonal contraceptive method than when not (for all methods except the patch), that frequency actually increases after adopting a hormonal contraceptive method, and that it remains higher than it was before adopting the hormonal method for several months, at least for oral contraceptives, the contraceptive ring, and LARCs (IUD and implant). This suggests that some other characteristic, such as relationship duration, may explain cross-sectional differences in sexual frequency among hormonal and non-hormonal contraception users. Contracepting couples increasingly switch from condoms to hormonal methods as their relationships endure, likely due to condom fatigue (Kusunoki & Barber, 2020) or increased trust and commitment to sexual exclusivity (Lehmiller et al., 2014; Manlove et al., 2014; Milhausen et al., 2018), and therefore accompany reduced fears of contracting a sexually transmitted infection from a partner. Thus, long-term couples are likely over-represented among any hormonal contraception users and therefore cross-sectional differences should be interpreted with caution.

Intimate relationships are influential contexts underlying women’s sexuality. Understanding hormonal contraceptive use in the context of relationships is critical, yet also presents unique challenges. For instance, some women’s hormonal contraceptive use extends beyond and/or across multiple intimate relationships (~13% of hormonal contraception episodes in the current study). In these cases, the women’s use of hormonal contraception is not dependent upon any specific relationship. This is consistent with research indicating several non-reproductive benefits and motivations for using these methods (e.g., health-related issues such as endometriosis, dysmenorrhea, or acne; Caruso et al., 2021; Frost & Lindberg, 2013), or could also indicate mere habit of use or concerns about any physiological changes following discontinuation (e.g., delayed or irregular periods, Nassaralla et al., 2011). Most episodes of hormonal contraception use in the current study, however, began and ended within the context of a single intimate relationship (or were still ongoing within that relationship when the study ended). This suggests much contraception decision-making occurs on the relationship level, or as a result of a specific relationship context. Understanding exactly how such contexts uniquely influence contraceptive decisions is an important next step.

The models also accounted for condom use and several relationship factors. Condom use was a potential confound that may have reduced fears of pregnancy (i.e., the proposed psychological mechanisms) as well as increased sexual frequency. Littlejohn (2021) demonstrated that many women were unwilling to have sexual intercourse or use hormonal contraception if their male partners were unwilling to use condoms, in an effort to equalize the gendered labor of contraception. Indeed, women who used condoms more frequently during the study period had more frequent intercourse, regardless of whether they used hormonal contraception. Our model showing that sexual frequency declines as relationship frequency increases echoes others who previously established this association using longitudinal or daily diary research methods (James, 1981; Jasso, 1985; Rao & Demaris, 1995; Schröder & Schmiedeberg, 2015). Our finding that sex was more frequent as relationship commitment (as evidenced by relationship type) increased is in contrast to one study demonstrating no effect of relationship type on sexual frequency (Schröder & Schmiedeberg, 2015). Our results may differ due to our focus on a broader range of relationship types (beyond cohabitation and marriage) or our focus on younger respondents. However, considering the full range of intimate relationships in which sexual intercourse occurs among young people, more frequent intercourse occurs in the more versus less committed types. Larger age differences between partners were also associated with more frequent sexual intercourse, which may reflect an imbalanced power dynamic, and is consistent with other research demonstrating similar associations (Kaestle et al., 2002).

Because long-term contraceptive users may differ from those who discontinue use more quickly—such as by being highly satisfied with their method or at least satisfied enough to continue use (Higgins & Smith, 2016; Sanders et al., 2001)—overall duration of contraceptive episodes was accounted for in two main ways. First, the multilevel models included average duration of episodes at the relationship and woman levels. Second, we conducted a separate analysis with the subsample of women who reported long-term (12+ months) use. Those analyses indicated no substantial differences in the associations for long-term versus short-term users. The indicator variable approach to the duration models minimized the potential confounding effects as well. The indicators also offered greater flexibility and precision with modeling the non-linear associations with sexual frequency in the data over time, particularly with the sharp increase immediately following initiation and the subsequent gradual changes evidenced by most methods.

Limitations and Future Directions

The RDSL was designed specifically to assess women’s experiences with sexual intercourse due to its inherent connection to reproductive and contraceptive experiences. There are of course many activities available for couples to foster sexual intimacy that were not captured in the RDSL study. Any decline in sexual intercourse may have been supplemented by an increase in oral sex, anal sex, or other non-penetrative activities. Future research assessing how contraceptive methods relate to a multitude of partnered sexual behaviors will help illuminate how specific methods influence partners’ sexual connections. Relatedly, partner gender was not assessed, which precluded our ability to focus solely on relationships in which penile-vaginal intercourse was possible. A supplementary survey administered to 590 of the women indicated that the majority (82%) identified as heterosexual (“straight”) (Weitzman et al., 2020), and that even the women who were not exclusively heterosexual typically had sex with one or more men at some point during the study, suggesting the majority of RDSL relationships were with male partners. The current study could also not assess sexual motivations, such as whether intercourse occurred because of the woman’s sexual interest or because she engaged in sex despite a lack of interest. Assessing couples’ motivations for engaging in sexual intercourse (or not) could better explicate mechanisms between hormonal contraceptive use and sexual behaviors.

Our study was limited to women in their early reproductive years. While this was a strength because of the relevance to the study aims, it does limit generalizability. Any conclusions should not be generalized to older women, who may be more or less prone or sensitive to negative sexual side effects or other sexual dysfunctions/dissatisfactions, or to women in very long-term relationships (e.g., 30-year marriages). Indeed, some evidence indicates the negative sexual side effects from hormonal contraception use, specifically declining sexual desire, may be more pronounced among older women (Schaffir, 2006).

The different method types were not used equally by these young women. Oral contraceptives were by far the most common (~80% of hormonal contraception reports), suggesting the combined analyses are more representative of this method than of other methods. Indeed, the separate exploratory analyses indicated a somewhat different trajectory of sexual frequency for the second most common method, the contraceptive injectable.

Menstruation was not accounted for, which could have affected the results, as some couples avoid sexual intercourse on bleeding days (Roney & Simmons, 2013), and some women’s hormonal contraception use may decrease periods of bleeding (e.g., using oral contraceptives continuously; Oxfeldt et al., 2020). It is unclear whether or how women’s menstruation experiences may have impacted the results, which presents an important direction for future research. Further, duration of hormonal contraception use was determined since the beginning of the study, and therefore did not account for duration prior to the study. The majority of the women (62%) were not using a hormonal method at baseline, and the majority of those who were (60%) discontinued that method within the first two months. Thus, most of the hormonal contraception use reported by the women began during the study. Given the young age of the women at baseline, and that young women may be more likely to discontinue use of a hormonal method, the small number of women who were using the method at baseline are unlikely to have been using the method for great lengths of time. This, however, cannot be confirmed in the current study. Researchers should account for this in future naturalistic studies of hormonal contraception use in young women’s everyday lives.

Conclusion

Helping couples who want to avoid pregnancy achieve those reproductive goals while also supporting their attempts to foster intimacy via sexual contact is important but challenging given the potential for negative sexual side effects from the most effective contraceptive methods. However, our analysis suggests that, for many method types, if any negative sexual side effects arose, the reduced fear of unintended pregnancy or potential positive sexual side effects may have played a larger role in sexual decision-making, at least for reproductive sexual behaviors (i.e., intercourse) among young couples, and at least in the moderate-term time horizon of two years. Despite sexual intercourse frequency remaining relatively high over time for users of some methods, the eventual declines in sexual frequency associated with some continued hormonal contraceptive use, combined with many couples’ reliance on hormonal contraception over many years, suggest it should remain a high priority to develop new hormonal contraceptive methods with fewer sexual side effects or new contraceptive methods that do not rely on hormonal mechanisms to prevent pregnancy.