Longitudinal changes in sexual desire and attraction among women who started using the Natural Cycles app

, duration of HC use) or relationship characteristics (e.g., relationship length). Results revealed that both sexual desire and partner attraction, on average, increased across three months after beginning to use Natural Cycles. While the prediction that changes in sexual desire would co-occur with cycle resumption was supported, there was also evidence that libido continued to increase even after cycles resumed. Together, these results offer new insights into relationships between HC discontinuation and women's sexual psychology and lay the groundwork for future research exploring the mechanisms underlying these effects.

Many women experience sexual side effects, such as decreased libido, when taking hormonal contraceptives (HCs).However, little is known about the extent to which libido recovers after discontinuing HCs, nor about the timeframe in which recovery is expected to occur.Given that HCs suppress the activities of multiple endogenous hormones that regulate both the ovulatory cycle and women's sexual function, resumption of cycles should predict libido recovery.Here, using a combination of repeated and retrospective measures, we examined changes in sexual desire and partner attraction (among partnered women) across a three-month period in a sample of Natural Cycles users (Survey 1: n = 1596; Survey 2: n = 550) who recently discontinued HCs.We also tested whether changes in these outcomes coincided with resumption of the ovulatory cycle and whether they were associated with additional factors related to HC use (e.g., duration of HC use) or relationship characteristics (e.g., relationship length).Results revealed that both sexual desire and partner attraction, on average, increased across three months after beginning to use Natural Cycles.While the prediction that changes in sexual desire would cooccur with cycle resumption was supported, there was also evidence that libido continued to increase even after cycles resumed.Together, these results offer new insights into relationships between HC discontinuation and women's sexual psychology and lay the groundwork for future research exploring the mechanisms underlying these effects.
The recommended therapy for women experiencing HC-related female sexual dysfunction is typically discontinuation of HCs and use of an alternative method of contraception, such as condoms, the copper intrauterine device, or (when future fertility is not a concern) sterilization of either partner (Casado-Espada et al., 2019).However, if women comply with these recommendations, little guidance can be given to them about what to expect next.In particular, little is currently known about a) whether sexual desire and responsiveness increase when women discontinue HC treatment, b) what the timeline is within which women should expect to observe changes for the better, and c) whether these changes coincide with the return of a woman's ovulatory cycle.
The current research was designed to address this gap.Specifically, we sought to examine whetherand on what timelinewomen experience changes in desire and partner attraction after discontinuing HC treatment.We predicted that women would report experiencing increases in sexual desire and partner attraction since discontinuing HC use.Further, because decrements in sexual desire and partner attraction are thought to be driven, in part, by HC-induced suppression of the hypothalamic-pituitary-gonadal (HPG) axis, we predicted that increases in sexual desire and partner attraction would co-occur with the return of women's cycles (i.e., resumption of periods).Additionally, we explored relationships between changes in women's sexual desire and partner attraction after discontinuing HC use and the following factors: (a) the type of HC women were discontinuing (e.g., oral contraceptives vs. intrauterine devices), (b) the length of time women spent on HCs, and (c) the amount of time that has elapsed since HC discontinuation.Finally, we also examined whether changes in sexual desire and partner attraction upon HC discontinuation are moderated by relationship characteristics, including (d) relationship status, (e) whether women chose their romantic partners while on or off of HCs, and (f) among partnered women, relationship duration.Results seek to lend insights into the direction and magnitude of changes in sexual desire and partner attraction women experience after discontinuing HCs, as well as personbased factors that may moderate these relationships.

Sexual side effects of hormonal contraceptive use
While many women who use HCs are able to do so without experiencing sexual side effects (Burrows et al., 2012;Pastor et al., 2013), other women have a different experience.Several cross-sectional studies have found that women using HCs report less sexual arousal and pleasure than those who are naturally cycling (Davison et al., 2008;C.W. Wallwiener et al., 2010;M. Wallwiener et al., 2010) or using nonhormonal birth control (Boozalis et al., 2016;Mark et al., 2016;Smith et al., 2014).Similar results have been found in randomized experiments, with women who received a combined oral contraceptive reporting less sexual interest after three treatment cycles relative to those who received a placebo, although the size of this effect was modest (Lundin et al., 2018).These negative side effects can have downstream consequences on women's sexual behavior and responsivity.For example, women using HCs tend to have sex less often than naturally cycling women and report more discomfort from sexual activity when they do have sex (Caruso et al., 2004;Smith et al., 2014;C.W. Wallwiener et al., 2010;M. Wallwiener et al., 2010;Wallwiener et al., 2015).
However, several studies have also found that many women do not report changes in sexual desire and behavior in response to HC use.For example, some research finds that only a minority of women experience sexual side effects while using HCs (Both et al., 2019), and other studies have even reported improvements in women's sexual health during HC use (relative to when naturally cycling) (Alexander et al., 1990;Botzet et al., 2021;Caruso et al., 2004;Mark et al., 2016).Mixed findings from research examining relationships between HC use and women's sexual function highlight the need for additional well-powered studies on the topic generally, and in particular, studies that investigate the individualand HC-level factors that may help explain differences in women's experiences with HCs.
Although there are multiple potential mechanisms through which HC use could influence sexual desire and attraction, there are three contributors that have been well-characterized by existing research: changes in levels of estradiol resulting from suppression of the HPG axis (Panzer et al., 2006;Reape et al., 2008;Søeborg et al., 2014;Stanton et al., 2011), increased levels of sex hormone binding globulins (SHBG) that decrease levels of free testosterone (Coenen et al., 1996;Wiegratz et al., 1995), and changes in other hormones that regulate women's attraction and pair-bonding behaviors (Hwang et al., 1986;Ismail et al., 1998;Klipping et al., 2021;Mishell et al., 1977).These are reviewed below.
Much research finds that increases in estradiol play a role in both partner attraction (Roney and Simmons, 2016) and sexual desire (Jones et al., 2018;Roney and Simmons, 2013).For example, animal studies find that estradiol is central to female sexual function in a variety of mammals and removing the ovaries reliably diminishes mating behavior, an effect that is reversed upon administration of exogenous estradiol (Beach, 1947;Cappelletti and Wallen, 2016;Wallen, 1990).Estradiol's important role in sexual function has been further supported by the results of human research finding that increases in sexual motivation co-occur with rises in estradiol levels (and reductions in progesterone) across the menstrual cycle (Roney and Simmons, 2013).Further, the precipitous drop in estradiol that occurs at menopause is accompanied by steep decreases in women's sexual desire (Woods et al., 2010), and exogenous estrogen therapy is often effective in restoring sexual desire in postmenopausal women (Cappelletti and Wallen, 2016;Sarrel, 2000;Scavello et al., 2019).Because HCs inhibit ovulation and its corresponding periovulatory estradiol surge, it is reasoned that one contributor to sexual dysfunction among HC users is the low levels of estradiol observed among HC-taking women, which itself may be attributable to lack of normal cyclical variation in estradiol among women taking HCs (Mishell et al., 1972;Pluchino et al., 2009;Scheele et al., 2016).
Beyond impacting estradiol levels, suppressing the HPG axis also has implications for women's levels of testosterone, which plays its own role in regulating women's sexual motivation and arousal (Vale et al., 2017;Wåhlin-Jacobsen et al., 2017).Research finds that HC use attenuates the effects of endogenous testosterone through multiple pathways.First, HCs reduce ovarian and adrenal production of testosterone and other androgens (e.g., dehydroepiandrosterone sulfate [DHEAS]), resulting in lower circulating levels of these hormones (Fern et al., 1978;Kuhl et al., 1985;Madden et al., 1978;Zimmerman et al., 2014).HC use is also associated with increases in levels of SHBG (Coenen et al., 1996;Wiegratz et al., 1995), which leads to especially precipitous declines in free, biologically active testosterone (Fern et al., 1978;van der Vange et al., 1990;Zimmerman et al., 2014).Although there is a paucity of direct evidence that decreases in free or total testosterone play a causal role in HC-induced sexual dysfunction, some research has provided preliminary support for this possibility.For example, one prospective study found that greater reductions in free testosterone across the three months following HC initiation were associated with steeper decreases in the frequency of sexual thoughts during this period, but not enjoyment of sexual activity (Graham et al., 2007).Further, one randomized control trial found that administering DHEA alongside combined oral contraceptives, which restored free testosterone levels to those measured before HC administration, mitigated some sexual side effects of HCs (van Lunsen et al., 2018).Moreover, this study also reported that women whose free testosterone levels were higher during treatment had higher sexual arousal and desire.However, separate research did not find consistent relationships between levels of free or total testosterone and sexual desire among HC users (Alexander et al., 1990).Mixed findings in this domain highlight the need for further research on links between testosterone and sexual health among women using HCs.
In addition to their impact on women's levels of sex steroid hormones, research finds that HCs can also impact other neuroendocrinological systems that regulate sexual function and partner attraction.For example, abnormally high levels of prolactin are associated with sexual dysfunction in both men and women (Meston and Frohlich, 2000;Nappi et al., 2021).Research finds that certain forms of HCs stimulate prolactin release from the anterior pituitary, increasing serum levels of this hormone (Hwang et al., 1986;Ismail et al., 1998;Klipping et al., 2021;Mishell et al., 1977).In extreme cases, HC use can even increase women's risk for developing clinical hyperprolactinemia (Badawy et al., 1981), which is a disease state characterized by marked decreases in sexual desire, arousal, and satisfaction (Kadioglu et al., 2005;Knegtering et al., 2008).In addition to prolactin, HC use may also interfere with the actions of oxytocin, a hormone heavily involved in promoting pairbonding and affiliative behavior (Algoe et al., 2017;Bosch and Young, 2018;Feldman, 2012).One recent study found that HC users exhibit dysregulated responses to intranasal oxytocin in a pair boding context relative to what is observed among naturally cycling women (Scheele et al., 2016).
In addition to these mechanisms that help explain why HC use may negatively impact sexual function, there are also pathways through which HCs may improve sexual function.For example, women might experience increases in sexual motivation when they do not have to worry about an unplanned pregnancy.Moreover, among women with hormonal disorders for which HCs may be prescribed as treatment (e.g., polycystic ovary syndrome), there is evidence that use of HCs reduces sexual dysfunction (Steinberg Weiss et al., 2021).Even outside the context of health problems, there are substantial individual differences in hormonal profiles, as well as individuals' sensitivity to hormonal fluctuations (Soares and Zitek, 2008).Therefore, the same hormonal mechanisms through which HCs decrease libido in some women (e.g., changes to endogenous estradiol, testosterone) could lead to increases in other women, or even in the same woman at a different life stage (e.g., pre-vs.perimenopause) (Zhang et al., 2023).
Together, this body of research suggests that there are multiple pathways through which HC use may impact women's sexual function and partner attraction, some of which are tied to the functioning of the HPG-axis (e.g., those tied to levels of estradiol) and others that are not (e.g., hyperprolactinemia).Although it is widely assumed that processes related to sexual desire and partner attraction will improve once women's ovulatory cycles resume and women's production of endogenous sex hormones is no longer suppressed, this assumption has not been tested.Such a test is necessary, however, as there is evidence that some physiological changes initiated by the pill may persist beyond the return to function of the HPG axis.For example, research finds that SHBG levels in former users of HCs are still higher months after discontinuation compared to those of never-users (Panzer et al., 2006).Others find that among women who were on HCs for one year, prolactin levels remained elevated for at least six months after having discontinued use (Alvarez-Tutor et al., 1999).These findings suggest that the hormonal processes underlying women's sexual function and attraction may remain dysregulated for months after HC use ends, potentially prolonging improvements in libido and partner attraction beyond the time that the cycle returns.Understanding the degree to which women report noticing positive changes in libido and partner attraction upon resumption of the ovulatory cycle is therefore important, as it can provide insight into the mechanisms that contribute to sexual dysfunction while on HCs and can provide an easy-to-identify benchmark to help HC users gauge expectations about improvements in libido that may occur after discontinuing HCs.

The current research
The current research sought to build on previous work examining the links between HC use and sexual function by examining changes in sexual desire and partner attraction following resumption of the menstrual cycle in a sample of 1596 Natural Cycles users who had recently discontinued HCs.Using a combination of retrospective and longitudinal measures, we tested whether resumption of the cycle post-HC-as measured by the return of menstruation-would coincide with increases in each sexual desire and partner attraction.Further, we explored whether changes in sexual desire and attraction were moderated by a variety of factors that may impact the HPG axis return-to-function, such as duration of HC use and hormone delivery modality.Women who have used HCs longer, for example, may exhibit greater latency to the return of ovulation and therefore greater latency to improvements in sexual desire and partner attraction compared to women who have used HCs for a shorter duration of time.It is also possible the type of HC product that women were on may moderate the return of sexual desire, as research suggests that these factors may contribute to observed heterogeneity in women's experience of sexual and relationship side effects (Hill and Mengelkoch, 2023;Roumen, 2008;Sabatini and Cagiano, 2006).
The current research also examined whether relationship characteristics that are known to play a key role in sexual desire and partner attractionincluding relationship status, relationship duration, and HC status at the time of relationship formationwere related to sexual function and partner attraction after discontinuing HCs.Research generally finds, for example, that women's sexual attraction to their partners wanes as relationship duration increases (Murray and Milhausen, 2012;van Lankveld et al., 2018).Others find that women's HC use status at the time they chose their partners may likewise impact sexual desire and partner attraction, particularly once HC use is discontinued (French and Meltzer, 2020;Roberts et al., 2012;Russell et al., 2014).Specifically, this research finds that women who chose their partners while using HC may have an attenuated return to desire after discontinuing HCs if there is incongruence between their partner preferences while on-versus off-HCs (for research supporting this hypothesis, see French and Meltzer, 2020;Roberts et al., 2012;Russell et al., 2014; for research failing to find evidence for this pattern see Jern et al., 2018).Accordingly, we tested whether these relationship characteristic variables moderated the predicted increases in each sexual desire and partner attraction following HC discontinuation.

Natural cycles application
Natural Cycles is the company behind the first Food and Drug Administration (FDA) cleared and CE-marked mobile application for digital contraception.The application estimates women's ovulation and identifies its corresponding fertile window using their records of menstruation and daily basal body temperatures.Depending on the user's intent to avoid or plan a pregnancy, these daily fertility predictions guide women in their decisions about sexual intercourse and protection through the use of a color scheme and corresponding instruction set.For example, for women using Natural Cycles for birth control, women's fertile days are shown in red and they are told to use protection if engaging in sexual activity.Women's non-fertile days are shown in green and women are told that they are not fertile (see Fig. 1).
Natural Cycles currently has over 2 million registered users.Users give consent within the app for their anonymized data to be used in scientific research.Consent can be withdrawn within the app at any time.Among the users who had given consent to participate in scientific research, a Natural Cycles research team blind to the purposes of the current study defined a cohort of users located in the United States who had recently discontinued HCs (within the last year).Participants were eligible for participation if they had not yet resumed menstruation or, if they had, they had not had more than two cycles.These users were specified to achieve a sample of women who had recently discontinued HCs, some of whom had resumed cycling and some of whom had not.Users were excluded from this cohort if they indicated a pregnancy in the year prior to registration.The chosen cohort was invited to participate through an in-application message that asked them to fill out a first questionnaire within 7 to 10 days after registration.If users replied and remained active within the Natural Cycles app, a second in-application message prompted them to answer a second follow-up questionnaire three months after registration.

Participants
Participants enrolled in the current research met the following inclusion criteria: (a) consented for their data to be used for science, (b) consented to receiving e-mails, (c) registered in the United States, (d) had quit HCs in within a year of data collection, but had had no more or than three menstrual cycles post-HC use at onboarding, (e) had not been pregnant in the year preceding participation, and (f) had at least one cycle after onboarding.On average, the time between discontinuing HCs and completing Survey 1 was 21.34 days (SD = 35.21).After duplicate entries were removed, a total of 1596 women completed the first survey, with 550 continuing to complete the second survey three months later (see Table 1 for sample characteristics).Multivariate analyses of variance (MANOVAs) revealed that participants who complete only Survey 1 did not significantly differ from those who completed both surveys by Survey 1 levels of sexual desire, partner attraction, or demographic characteristics (see Table S3 for results).

Longitudinal changes in libido and attraction
To measure longitudinal changes in sexual desire across the study period, participants responded to the questions: (a) "How would you rate your sexual desire today?" (1: extremely low levels of sexual desire, 4: neither low nor high levels of sexual desire, 7: extremely high levels of sexual desire) and (b) "How would you rate your sexual desire in general, over the last few weeks?" (1: extremely low levels of sexual desire, 4: neither low nor high levels of sexual desire, 7: extremely high levels of sexual desire).Both questions were asked at the first and second surveys.The following question was also responded to at both surveys to examine longitudinal changes in sexual attraction to one's partner: "How much sexual attraction do you feel toward your current partner?" (1: extremely low levels of sexual attraction, 4: neither low nor high levels of sexual attraction, 7: extremely high levels of sexual attraction).

Perceived changes in libido and attraction
To examine participants' perceived changes in sexual desire and partner attraction since discontinuing HCs, participants who responded to the second survey answered the retrospective questions: (a) "Have you noticed any changes in your sexual desire since discontinuing birth control?" (1: much lower levels of sexual desire, 4: neither lower nor higher levels of sexual desire, 7: much higher levels of sexual desire) and (b) "Have you noticed any changes in your feelings of attraction to your partner since discontinuing hormonal birth control?" (1: much lower levels of sexual attraction, 4: neither lower nor higher levels of sexual attraction, 7: much higher levels of sexual attraction).These items were measured at Time 2 only to ensure that all women were able to reflect on at least three months of self-perceptions.

Type and duration of HCs
Participants also provided information about the specifics of their HC use that could potentially impact levels of sexual desire and partner attraction.Specifically, participants responded to the question, "What hormonal birth control were you using right before switching to Natural Cycles?".Participants responded by selecting (a) combined oral contraceptive pill ("the pill"), (b) hormonal IUD, (c) hormone releasing vaginal ring, (d) progestin-only oral contraceptive pill ("the mini pill"),  Note.IUD = intrauterine device.a For 64 women, data on cycle resumption between surveys was not available.
(e) the implant, (f) the patch, (g) the shot, or (h) other (with text entry), and then indicating the brand name.Responses were collapsed into a binary variable comparing oral contraceptives (combined and progestinonly oral contraceptive pills; coded 1) to all other non-oral products (coded 0).Among women who took an oral contraceptive pill, the generation of the progestin included in each formulation was determined based on brand names provided by participants in response to a follow-up question (coded generations 1-4) (Cooper et al., 2022).A follow-up analysis compared combined oral contraceptive pills to progestin-only pills.
At the first survey, participants were asked to provide an approximate start date on which they began HC use prior to switching to Natural Cycles. 1 To measure cycle resumption, women were asked in the application whether they had a period since discontinuing HCs, and if so, to provide the date of their last period (since stopping HCs).Women who had started their periods by study enrollment were asked to list the number of menstrual cycles they had since discontinuing HCs (range: 1-3).Users' daily cycle tracking data between Survey 1 and Survey 2 were used to determine whether women who had not yet resumed cycling at the time of Survey 1 had done so by the time they completed Survey 2 (i.e., if a period was reported), as well as to determine how many cycles were logged between the two surveys.For all analyses using these data, we tested (a) whether the outcome differed depending on whether women had yet resumed cycling after discontinuing HCs (i.e., binary variable; periods had or had not returned) and (b) whether the number of cycles women reported having since HC discontinuation predicted outcomes (among women who had resumed menstruating).

Relationship characteristics
Across the two surveys, information about participants' relationship status was collected to examine whether relationship characteristics moderated the extent to which sexual desire and partner attraction changed following HC discontinuation.These included whether or not the participant was in a romantic relationship (Surveys 1 & 2; binary), the length of that relationship (Survey 1; calculated as days based on dates provided by the participant), whether the participant met her partner while on or off HCs (Survey 1; binary), and whether the participant's relationship status had changed over the course of the study (Survey 2; binary).All models analyzing relationship data were restricted to women who were in a relationship at Survey 1, and whose relationship status did not change by the time they took Survey 2. (i.e., only women who were in the same relationship across the study; n = 469).

Alternative explanations
Data were collected on a number of variables that could offer alternative explanations for the hypothesized relationship between HC discontinuation, cycle resumption, and changes in sexual desire.These included age, highest level of education completed (1 = elementary school, 2 = high school degree, 3 = trade/technical/vocational training, 4 = university degree, 5 = PhD), number of previous pregnancies (0 = none, 1 = one, 2 = two, 3 = three or more), and reason for using Natural Cycles (henceforth referred to as "goal").Regarding the latter, women reported whether they were primarily using Natural Cycles to either prevent (coded 0) or plan (coded 1) a pregnancy.

Data analysis plan
Data analysis was pre-registered at doi:10.17605/OSF.IO/YV7F2.Although no analyses were conducted on the data prior to publishing the pre-registration, data were received and converted into tables before the registration was submitted.See Tables S1-S2 for a list of tested predictions.The registration document describes hypotheses and predictions for both survey data and mobile application data.Only the survey data were analyzed for the current research (with the exception of data on cycle resumption between the surveys); results for application data are forthcoming in a separate manuscript.All analyses were conducted using SPSS (v26) and MPlus (v8) statistical software; analysis code is available at https://osf.io/wpjda/.Relationship length and time spent on HCs were positively skewed and log-transformed prior to analyses.The sexual desire and partner attraction variables were normally distributed at both timepoints.These data were analyzed using latent change score models (see Fig. 2 for diagram), which allow us to test for changes in outcomes measured repeatedly in a study.These models hold a number of advantages over other types of change score and repeatedmeasures analyses (Coman et al., 2013;Kievit et al., 2018;Klopack and Wickrama, 2020).For example, latent change score modeling allowed us to treat sexual desire at each timepoint as a latent factor, represented by the two observed variables related to this construct that were collected at each timepoint (sexual desire today, sexual desire in last few weeks).Modeling sexual desire as a latent factor helps account for measurement error and the reality that any singular variable provides an incomplete picture of the overall construct (Bollen, 2002;Llabre and Fitzpatrick, 2012;McArdle, 2009).Further, latent change score models offer a flexible way to track changes in variables across time while (a) modeling the relationship between the magnitude of change and initial levels (here, Survey 1 levels of each variable) and (b) allowing for the addition of continuous and categorical predictors of both initial levels and change.
Both initial levels of sexual desire and partner attraction, as well as the latent changes in each construct over time, were regressed on the HC use-and relationship-level predictors.Model fit for the latent change score models was assessed using the following indices: (a) χ 2 test of model fit, (b) the comparative fit index (CFI), (c) the root mean square error of approximation (RMSEA), and (d) the standardized root mean square residual (SRMR).Good model fit was determined by a nonsignificant χ 2 value (p > 0.05), a CFI value >0.95, an RMSEA value <0.08, and an SRMR statistic <0.08.Because χ 2 values can be inflated with large sample sizes (Babyak and Green, 2010), models that yielded a significant χ 2 value, but for which all other fit indices were within range, were considered acceptable.We also examined whether the number of cycles reported predicted each Survey 1 levels of desire/attraction (i.e., cycles prior to the first survey) and Survey 2 levels (i.e., all cycles prior to second survey).These additional analyses were included to specifically probe whether the return of HPG function (indicated here by cycle resumption) influences libido and attraction in an all-or-nothing fashion, or whether instead the accumulation of more cycles after periods resume has an additive effect on sexual desire.
Perceived changes in sexual desire and partner attraction (Survey 2 only) were both negatively skewed, but because the units of analysis for these data were the frequency of increases in each relative to scale midpoints (i.e., no change), the variables were not transformed.Onesample t-tests were used to compare the group mean of each measure to the scale's midpoint.In follow-up models, perceived changes in sexual desire and partner attraction were regressed on the aforementioned predictors, with the exception of progestin generation.Because progestin generation was a four-level categorical variable, its relationship with each target outcome was tested using univariate analyses of variance (ANOVAs).The results of these analyses are available in the Supplementary materials.
For each set of outcomes (those measured longitudinally, as well as those measured at Time 2 only), a series of follow-up models was tested to examine possible alternative explanations for the expected pattern of results.In each of these models, we controlled for the following covariates: age, education, number of previous pregnancies, and goal (i.e., whether women were trying to plan or prevent a pregnancy).Examining the effects of women's goals was of particular importance given the impact of expectancies of intimacy on women's sexual arousal 1 Recruitment was targeted toward women who had recently discontinued HC use.(Blumenstock, 2022).Covariates were included in secondary models only because their inclusion reduced statistical power due to missing data in initial models (see Table 1 for variable-specific sample sizes).Effects of covariates on the target outcomes are reported in text as follow-up models; effects for focal predictors while controlling for covariates are reported in the Supplementary materials.An additional set of follow-up models excluded women seeking to become pregnant as an additional check on the possibility that app use intention was driving the observed changes in desire and attraction over time.
A final set of follow-up analyses examined whether there were any theoretically meaningful differences between women whose sexual desire and partner attraction decreased, stayed the same, or increased across the duration of the study.Specifically, we tested whether these groups of women differed with respect to demographics, relationship status, or HC use history.While the primary analyses were centered around the relationship between the key predictors and degree of change in attraction and libido, these follow-up models focused on the frequency of negative and positive changes in these variables experienced by women in the sample, regardless of magnitude.These analyses complemented the primary results by offering additional insights into the consistency of increases or decreases in libido and attraction after resumption of cycling, as well as the extent to which the direction of change was related to variables of interest.Participants were classified into a one of three levels in our categorical variable (decreasing, staying the same, or increasing) for each sexual desire (in recent weeks) and partner attraction based on the differences between Survey 1 and Survey 2 responses; differences between categories in demographic, relationship, and HC use variables were analyzed using one-way ANOVAs.

Longitudinal changes in sexual desire after discontinuing HCs
Characteristics of the sample are shown in Table 1.See Table 2 for descriptive statistics.Sexual desire was modeled as a latent factor consisting of responses to two questions asked at each survey: (a) "How would you rate your sexual desire today?" and (b) "How would you rate your sexual desire, in general, over the last few weeks?".Correlations between the two items were r = 0.67 at Survey 1 and r = 0.65 at Survey 2. Results of the latent change score model (see Fig. 2) revealed good model fit, χ 2 (3) = 43.77,p < 0.001; CFI = 0.96; RMSEA = 0.07; SRMR = 0.05.There was a modest, but statistically significant increase in reported sexual desire from the first to second survey (β = 0.15, SE = 0.07, t = 2.19, p = 0.03).Notably, baseline levels of sexual desire and the change from the first to second survey were strongly negatively correlated (β = − 0.62, SE = 0.04, t = − 17.01, p < 0.001), suggesting that the  Note.M = mean, SD = standard deviation, SE = standard error.
J. Gassen et al. magnitude of this increase was much smaller for women whose levels of sexual desire were relatively high at the beginning of the study.

Effects of cycle resumption on sexual desire
Results of the latent change score models are summarized in Table 3. Results revealed that women whose menstrual cycles had resumed by Survey 1 reported higher levels of sexual desire at this timepoint than those whose cycles had not yet resumed (β = 0.12, SE = 0.03, t = 3.73, p < 0.001).Further, among women who had not resumed cycling by Survey 1, but whose cycles had returned by Survey 2 (i.e., women who experienced a change in cycle resumption status between the two surveys; n = 294), 2 there was a significant increase in sexual desire across the study that coincided with cycle resumption (β = 0.21, SE = 0.11, t = 1.97, p = 0.049).While the effect size for this sub-group was larger than for the sample as a whole (i.e., β = 0.15), there was also a high standard error (see Discussion for more information).Although women whose cycles had already returned by Survey 1 also exhibited, on average, an increase in sexual desire from the first to second survey (β = 0.12, p = 0.36), this effect did not reach statistical significance.

Relationship between cycle number and sexual desire at each timepoint
Next, to test whether there is an additive effect of cycle number on libido above the influence of cycle resumption alone, we examined relationships between cycle number and sexual desire at each time point.Among women whose cycles had returned by Survey 1, a greater number of cycles since discontinuing HCs (i.e., prior to Survey 1) was associated with higher levels of sexual desire at this timepoint (β = 0.17, SE = 0.05, t = 3.60, p < 0.001).However, for all women whose cycles returned by Survey 2, total cycle number (i.e., after discontinuing HC use) was not related to sexual desire at the second survey (β = 0.04, p = 0.41).

Effects of type and duration of HC use on sexual desire
Results further revealed that women who took an oral contraceptive pill (combined or progestin-only) had lower initial levels of sexual desire than women who used other hormone delivery modalities (β = − 0.06, SE = 0.03, t = − 2.18, p = 0.03), as well as a greater increase in sexual desire from the first to second survey (β = 0.11, SE = 0.05, t = 2.43, p = 0.02).Among women who took an oral contraceptive pill, progestin generation was not consistently related to either starting levels of sexual desire, or the change in sexual desire over time (ps: = 0.88-0.97).
Duration of HC use was related to both initial levels of sexual desire (β = − 0.12, SE = 0.05, t = − 2.74, p = 0.006), and the change in desire from the first to last survey (β = 0.15, SE = 0.08, t = 2.03, p = 0.04).Specifically, women who reported being on HCs longer reported lower sexual desire at the beginning of the study, but a greater increase over time than women who were on HCs for a shorter period of time.The size of the effect of HC use duration was similar to that of change in desire from the first to second survey for the sample as a whole.

Effects of relationship status, relationship length, and HC status at relationship formation on sexual desire
Neither initial levels of sexual desire (β = 0.003, p = 0.48), nor the change in sexual desire over time were associated with relationship status (β = 0.02, p = 0.46).However, longer relationship length (among women in a relationship) was related to lower starting levels of sexual desire (β = − 0.22, SE = 0.03, t = − 7.92, p < 0.001) and a greater increase in sexual desire over time (β = 0.16, SE = 0.05, t = 3.10, p = 0.002).Furthermore, baseline levels of sexual desire were not associated with HC status at the time of relationship formation (among women in relationships; β = 0.02, p = 0.74).While women who met their partners off of HCs tended to report a greater increase in sexual desire over time relative to those who met their partners on HCs, this relationship did not reach statistical significance (β = − 0.07, p = 0.26).

Longitudinal changes in partner attraction after discontinuing HCs
Among women who were in a relationship, the change in partner attraction from the first to second survey was expressed as a latent change score of responses to the question, "How much sexual attraction do you feel toward your current partner?", asked at each timepoint.This model was just-identified, so fit statistics were not available.Results revealed that partner attraction increased over time (β = 0.24, SE = 0.04, t = 5.46, p < 0.001).The size of this effect was larger than that of the change in sexual desire (i.e., β = 0.15).As with sexual desire, women reporting higher initial levels of partner attraction exhibited a smaller increase in attraction across the study (β = − 0.65, SE = 0.03, t = − 21.31, p < 0.001).The magnitudes of change in sexual desire and partner attraction over time were positively correlated (β = 0.30, SE = 0.07, t = 4.34, p < 0.001).

Effects of cycle resumption on partner attraction
Results revealed that whether a participant's menstrual cycle had returned by Survey 1 was not related to partner attraction at this timepoint (β = 0.08, p = 0.09).For both women whose cycles had resumed by Survey 1 (β = 0.23, SE = 0.08, t = 2.82, p = 0.005), and women whose cycles had not returned by the first survey, but did resume by Survey 2 (β = 0.23, SE = 0.07, t = 3.46, p = 0.001, n = 261), there were significant increases in partner attraction across the study and with effect sizes of similar magnitude to that of change in the sample overall.

Relationship between cycle number and partner attraction at each timepoint
Among women whose cycles had returned by Survey 1, as with sexual desire, a greater number of cycles since discontinuing HCs (i.e., prior to Survey 1) was related to higher levels of partner attraction at this timepoint (β = 0.13, SE = 0.05, t = 2.79, p = 0.005).In contrast to the results for sexual desire, however, a greater number of total cycles was also associated with a small increase in partner attraction at Survey 2 (β = 0.09, SE = 0.04, t = 2.04, p = 0.04).

Effects of type and duration of HC use on partner attraction
Neither initial levels of partner attraction (β = − 0.02, p = 0.64), nor 2 The overwhelming majority of women reported that their periods had returned prior by the time they completed Survey 2 (97.2 %), including 95.9 % of women whose cycles had not resumed by Survey 1.
the change in partner attraction (β = 0.07, p = 0.09), were related to hormone delivery modality.Progestin generation for women who took an oral contraceptive pill was also unrelated to starting levels of partner attraction or the change in attraction over time (ps: = 0.58-0.81).
Comparing women who took a progestin-only pill with those who took a combined pill did not yield significant results (initial levels: β = 0.03, p = 0.50; change: β = − 0.08, p = 0.07).Length of time on HCs was not associated with starting levels (β = − 0.10, p = 0.14) nor the change in partner attraction over time (β = 0.10, p = 0.14).

Effects of relationship length and HC status at relationship formation on partner attraction
Results revealed that longer relationship length was associated with lower levels of partner attraction at the time of the first survey (β = − 0.20, SE = 0.05, t = − 4.23, p < 0.001), but a greater increase in attraction over time (β = 0.11, SE = 0.05, t = 2.35, p = 0.02).Further, women whose relationships began when they were off HCs (compared to those who were on HCs) reported greater initial levels of sexual attraction to their partners (β = − 0.11, SE = 0.05, t = − 2.33, p = 0.02), but HC status at relationship formation was not related to the change in attraction from the first to second survey (β = 0.02, p = 0.74).
In addition to controlling for app use goal, we also conducted followup analyses excluding women seeking to become pregnant, as it is possible that changes in desire and attraction would be greater for these women.As with the primary model, there were modest, but statistically significant increases in reported sexual desire (β = 0.19, SE = 0.08, t = 2.25, p = 0.03) and partner attraction (β = 0.23, SE = 0.05, t = 4.68, p < 0.001) from the first to second survey.

Follow-up change frequency analyses
The final follow-up analysis examined sample frequencies of change (of any magnitude) in sexual desire and partner attraction from Survey 1 to Survey 2 (see Table 4).Categories of change were simply determined by subtracting Survey 1 scores from Survey 2 scores; differences of 0 were categorized as 'no change', positive scores were categorized as 'increased,' and negative scores were categorized as 'decreased.'For sexual desire, the majority of women exhibited an increase over time (44.8 %), with nearly equal numbers of women reporting no change (27.1 %) or a decrease (28.1 %).As was found with the latent change score model, the participants who were most likely to experience an increase were those reporting low levels of sexual desire at the time of the first survey.Women who had more recently discontinued HCs were also more likely to observe an increase in sexual desire, but again, this was because their levels of desire were low at Survey 1.
While more women reported increases (36.1 %) than decreases (21.6 %) in partner attraction across the study, reporting no change in attraction was the most frequent response pattern (42.3 %).As with sexual desire, participants who increased in partner attraction tended to have low levels of attraction at Survey 1. Further, women who took a progestin-only oral contraceptive pill were more likely to report a decrease in partner attraction from Time 1 to Time 2 relative to those who took a combined oral product.No other sample characteristics were related to the likelihood of change.

Results for retrospective sexual desire and partner attraction outcomes
Results revealed that, consistent with the repeated-measures analyses, women reported noticing that their sexual desire increased after HC discontinuation, M = 5.39, SD = 1.24, t(453) = 23.82,p < 0.001, d = 1.12.Similarly, on average, partnered women reported feeling more attracted to their partners after discontinuing HCs, M = 5.00, SD = 1.10, t(401) = 18.30, p < 0.001, d = 0.91.In contrast to the results of the longitudinal analyses, perceived changes in sexual and desire and partner attraction were found to be largely unrelated to type of HC used, length of time on HCs, and relationship characteristics (e.g., length of relationship).

General discussion
In the current work, we investigated changes in sexual desire and partner attraction following discontinuation of HCs in a large sample of Natural Cycles users, using both repeated-measures and retrospective methods.Based on past research which finds some women report negative sexual side effects of HC use (Boozalis et al., 2016;Coelho and Barros, 2019;Novick et al., 2022;Sanders et al., 2001;C.W. Wallwiener et al., 2010;M. Wallwiener et al., 2010), as well as research finding that women may report changes in their attraction to their partners when changing HC use (French and Meltzer, 2020;Roberts et al., 2012;Russell et al., 2014), we predicted that after HC discontinuation, women would perceive increases in sexual desire and, for women in romantic relationships, increased partner attraction.Moreover, because suppression of the HPG axis likely contributes to the observed decrements in sexual desire often reported by women using HCs, we predicted that the return of libido and partner attraction would coincide with the resumption of women's cycles, which was expected to occur in most women within three months (i.e., by the time Survey 2 was administered).

Primary results
The results of the current research found support for some of our predictions.Women experiencedand perceived experiencingincreases in their sexual desire and attraction across the study.Moreover, consistent with the prediction that positive changes in women's sexual desire would coincide with the return of the ovulatory cycle, we found that women whose cycles had returned by the first survey reported higher levels of sexual desire at that time compared to women who had not yet resumed cycling.Further, among the sample of women who were not cycling at Survey 1, the return of cycles within the three months between Survey 1 and Survey 2 was accompanied by a significant increase in sexual desire.This pattern of results provides preliminary evidence that changes in libido in the months following HC discontinuation may coincide with normalization of ovarian function and the return of menstrual cycles.
We also found that among women who had started cycling prior to Survey 1, a greater number of reported cycles predicted higher levels of sexual desire at this timepoint.These results may suggest that some women experience additional gains in libido over the first few cycles after cycling resumes.In other words, the return of libido following HC discontinuation may not occur in an all-or-nothing fashion alongside cycle resumption, but rather libido continues to increase after this point, at least for some time.We did not find significant relationships between total number of cycles reported prior to Survey 2 and sexual desire at this later timepoint, which likely suggests that further increases in libido after cycling resumes may level off after the first few cycles.That is, in contrast to Survey 1 when most women reported fewer than three cycles, by Survey 2, virtually all women had resumed cycling and 80 % had recorded at least three cycles.That significant relationships between cycle number and sexual desire were found when most women reported 1-3 cycles (Survey 1), but not when most reported 3-8 cycles (Survey 2), provides evidence for a ceiling of post-HC libido recovery around the third cycle after periods return.Moreover, although there was a significant increase in sexual desire across the study among women who had not resumed cycling by Survey 1, but did resume by Survey 2, this effect carried a high standard error relative to the effect for the sample as a whole.This could be due to the smaller sample size for this sub-analysis, but it could also reflect heterogeneity in reasons why it took longer for women in this group to resume cycling.For example, a delay in recovery of normal cycling could be due to relatively high suppression of the HPG axis by HC use, but could also be due to stress, diet, or a myriad of other reasons.
Our prediction that increases in partner attraction occurring after discontinuing HCs would also coincide with the resumption of the cycle was not supported.Although women overall reported experiencing an increase in partner attraction across the study, these changes did not vary depending on women's cycle resumption status (cycle resumption: yes, no).When coupled with the results on changes in sexual desire coinciding with resumption of cycling, this result suggests that changes in partner attraction that can occur for women on HCs (Roberts et al., 2014(Roberts et al., , 2012) ) may have different or additional mechanistic underpinnings than changes in desire.It is possible, for example, that changes in partner attraction while on HCs are also influenced by oxytocin signaling, which has been reported in others' work (Scheele et al., 2016).Future research would benefit from investigating this, as well as other hormonal mechanisms to better understand how women's partner attraction changes when on or off of HCs.
While we found evidence that changes in desire and attraction coincided with the return of ovarian function following HC discontinuation, the nature of the data precludes making causal inferences about these effects.Most notably, initial measurements of sexual desire and partner attraction did not occur before or during HC use, but rather soon after they discontinued using HCs.Accordingly, we do not know if and how the target outcomes differed relative to levels before or while women were taking HCs.This serious limitation should be taken into account when interpreting the results of this research and considering their implications for sexual health (see 'Strengths, limitations, and future directions' for further discussion).

Moderators of changes in desire and partner attraction
Although previous research finds that many women experience decreased libido while using HCs, not all women share this experience (Boozalis et al., 2016;Coelho and Barros, 2019;Novick et al., 2022;Sanders et al., 2001;Smith et al., 2014;C.W. Wallwiener et al., 2010;M. Wallwiener et al., 2010;Wallwiener et al., 2015).Similarly, we found that there were individual differences in the presence and direction of changes to libido and partner attraction after HC discontinuation.Even though there was an overall increase in mean levels of sexual desire across the study, analysis of the frequency of different patterns of change (i.e., decrease, no change, increase) revealed that just under half of women experienced an increase in desire (44.80 %), while 28.10 % experienced a decrease, and 27.10 % reported no change.With partner attraction, 21.60 % reported a decrease, 36.10 % reported an increase, and 42.30 % reported no change.These results not only highlight heterogeneity in women's responses to HC discontinuation, but also individual differences in sexual function more broadly.The finding that the strongest predictors of change in desire and attraction were starting levels (i.e., at Survey 1) seems to suggest that a woman's sexual health prior to and during HC use may play important roles in the extent to which desire changes following HC discontinuation.In addition to the relationship and HC-use characteristics measured in the current work discussed below, there are a number of other biological and psychological factors that may influence sexual responses to HC discontinuation (see 'Potential mechanisms').We examined the effects of several variables that may lend insights into for whom and under which conditions discontinuation of HC use is related to changes in sexual desire and partner attraction.Here, we found that women who were using oral HC pills, women who have been using HCs for longer durations of time and (among women in relationships) those who were in longer relationships, reported: (a) lower levels of sexual desire immediately after ceasing HC use and (b) a larger increase in sexual desire after three months, compared to those using nonoral HCs, those who had been using HCs for a shorter duration of time, and those in newer relationships.That initial levels of sexual desire (i.e., soon after discontinuing HCs) were negatively related to the magnitude of change in sexual desire from the first to second survey may indicate that women whose libido is lowest soon after discontinuing HCs are the most likely to experience notable increases in sexual desire after ceasing HC use.This pattern could also merely reflect regression to the mean, which should be considered when interpreting these results.Further, findings also suggest that women using oral (compared to non-oral) HCs and those who have used HCs for a longer (compared to shorter) duration of time are more likely to follow this pattern, experiencing diminished libido soon after discontinuing HCs, but substantial recovery in the months following discontinuation.Although we did not directly measure libido prior to or during HC use, these results may suggest that (a) being on HCs (particularly oral products) for extended periods of time could increase women's risk of developing low libido, (b) recovery of libido for these women after discontinuing HCs is common, and (c) libido recovery may, in large part, occur within three months following HC discontinuation.
When it came to partner attraction, we found that women who had met their partners while not using HCs reported higher levels of partner attraction at Time 1 compared to women who had met their partners while using HCs.This result is consistent with the HC congruency hypothesis (Alvergne and Lummaa, 2010;French and Meltzer, 2020;Roberts et al., 2014;Russell et al., 2014), which suggests that because women's preferred attributes in a potential partner change when on or off HC, the preferences of a woman who discontinues HCs should align more closely with her preferences before ever starting HCs, compared to when taking HCs.However, it should be noted that previous studies on the congruency hypothesis have yielded mixed results (Botzet et al., 2021), and the effects of HC status at relationship formation on sexual desire and changes in partner attraction over time were not statistically significant.Additionally, we found that women in longer (compared to shorter) relationships reported lower initial partner attraction that increased more following HC discontinuation.Finally, as with sexual desire, there was a strong negative relationship between initial levels of partner attraction (Survey 1) and the change in attraction across the study.That is, women whose levels of sexual attraction toward their partners were lowest soon after discontinuing HCs experienced the greatest increases in attraction in the three months that followed.Unsurprisingly, the degree of change in partner attraction was also positively correlated to changes in sexual desire, indicating that these factors tend to shift together in direction and magnitude following HC discontinuation.Given the correlations between change in the constructs of general sexual desire and partner attraction, as well as the correlations between individual items comprising these constructs (rs: 0.49-0.58), it is important to keep in mind that there is considerable overlap between overall desire and specific attraction to a partner.The current research only employed a limited number of items to capture these constructs; future research using a broader range of measures is needed to more clearly delineate how each construct is uniquely affected by HC discontinuation.

Potential mechanisms
The results of the current study indicate that discontinuation of HC use is associated with changes in women's libido and attraction to their current partners.While this work was not designed to uncover the specific mechanisms by which HC use and its discontinuation are related to these variables, it is likely that these effects are driven by a number of factors, both biological and psychological.With respect to biological changes, HCs work to prevent pregnancy by suppressing the activities of the HPG axis, preventing ovulation,3 and in turn, pregnancy (Alexander et al., 1990;Mishell et al., 1972;Panzer et al., 2006).However, these changes also disrupt the release of endogenous estradiol, which can suppress libido given estradiol's links with sexual desire (Roney andSimmons, 2016, 2013) and sexual function (Caruso et al., 2004;Smith et al., 2014;C.W. Wallwiener et al., 2010;M. Wallwiener et al., 2010;Wallwiener et al., 2015).Further, HCs are known to be linked to significant increases in levels of SHBGs (Panzer et al., 2006;van der Vange et al., 1990), that bind to available sex steroids, including testosterone (Erodogan et al., 1980;Søeborg et al., 2014;Zimmerman et al., 2014), a hormone that also plays an important role in women's sexual arousal and libido (Vale et al., 2017;Wåhlin-Jacobsen et al., 2017).
Although research suggests that suppression of the HPG axis may persist for 2-6 months following discontinuation of HC use (Harlap and Baras, 1984;van den Berg et al., 2010), and there is some evidence that SHBG levels remain elevated in women who discontinue HC use for up to six months (Panzer et al., 2006), the current research suggests that the recovery timeline for functionally meaningful changes in these processes may occur much sooner.For example, in the current study, over 97 % of women reported at least one period by the end of the study, which provides evidence that resumption of HPG function in the three months following HC discontinuation is more the rule than the exception.In a subset of women for whom we had both data on date of HC discontinuation and date of first menstrual cycle post-HC use (n = 544), the mean time to cycle return was 35.41 days.However, the standard deviation was larger than the mean (SD = 43.88),indicating that there is a substantive amount of variation in how long it takes for women's HPG axes to normalize after HC discontinuation.Mixed findings in research examining the timeline of HPG regulation post-HC use may reflect this variation, and heterogeneity between study results could be attributable to differences in the sample characteristics found here to influence libido and attraction.Future research would benefit from examining this issue in greater depth, to provide women and their physicians a deeper understanding of the factors which influence the recovery timeline after HC use cessation.
Sexual desire before, during, and after HC use is also likely influenced by a host of psychological factors.For example, much research finds that many emotional and cognitive aspects of women's well-being have downstream effects on libido.Accordingly, women's mood and stress levelsboth of which can be adversely affected by HC use (Bengtsdotter et al., 2018;Ott et al., 2008;Kirschbaum et al., 1999;Lovallo et al., 2019) impact women's sexual proceptivity and receptivity (Bodenmann et al., 2010;Bodenmann and Ledermann, 2007;Nelson et al., 2008).It is therefore possible that the recovery of libido observed in the current research emerged in response to positive changes in mood and perceived stress, a possibility which should be explored in future research.
Finally, the results of the current research provide some evidence that mate choice may interact with the various biological, psychological, and cognitive effects of HCs to influence sexual desire and partner attraction.Specifically, consistent with the HC congruency hypothesis, we found that women who chose their partners off of HC tended to report higher levels of sexual attraction to that partner following HC discontinuation (compared to those whose relationships started while on HCs).Whether the results here are driven by HC-based shifts in mate preferences or alternative explanations requires confirmation in additional studies.For example, it is possible that women who meet their partners off HC have higher overall sexual desire at the time of relationship formation which translates into greater relationship satisfaction over time.

Strengths, limitations, and future directions
The current work had many notable strengths.By recruiting women who had just begun using the Natural Cycles application, we were able to collect a unique and large sample of women who had recently discontinued their HC use.We were also able to collect data from a subset of these women longitudinally, allowing us to investigate how their sexual desire and attraction to their current romantic partner changed over time.However, despite these strengths, there are also limitations to the current work that should be considered when interpreting the findings.
First, our sample of women was not drawn purely at random from the populationinstead, women self-selected to be a part of our study by both beginning to use the Natural Cycles application and by responding to our invitation to participate in research.Specifically, all of the women in our sample had chosen to discontinue their HC use and begin using the Natural Cycles application either as a natural form of contraception, or because they desired to become pregnant.Women in our study desiring to become pregnant may have been especially sexually motivated immediately after discontinuing HC use, and thus might report higher sexual desire as a result of this motivation to become pregnant once pregnancy from intercourse became possible (i.e., following HC cessation).Indeed, women using Natural Cycles to plan, as opposed as to prevent, a pregnancy reported higher sexual desire immediately following HC discontinuation.However, excluding women seeking to become pregnant from the primary analyses did not alter the pattern or significance of change in either libido or partner attraction across the study.Additionally, the women who began using Natural Cycles as a form of contraception might be women who experienced the most severe negative side effects as a result of their HC use, who are also most likely to benefit from a non-hormonal contraceptive option.In either case, effect sizes may be inflated in the current sample, and replication of these results in unique samples will be needed in the future.
Given that the current research was observational, there are a number of unmeasured confounds that could influence the results which should be considered in future research aiming to investigate changes in libido following HC discontinuation.With the current research, given that all women in the sample recently discontinued HCs, our primary models merely examined whether the latent changes in sexual desire and partner attraction that occurred from the first to second survey differed from 0. However, we infer in our interpretation that a similar sample of women who had not discontinued HCs (or who were not using HCs at all) would not exhibit similar changes over this period.Accordingly, confounds in this context would include variables that causally predict both HC use/discontinuation, as well as sexual desire and partner attraction.For example, seeking to become pregnant may prompt a woman to discontinue HCs and also change motivations for sexual activity (although controlling for app use goal did not influence the current results).Another potential confound could be sociosexual orientation prior to HC use, which might influence a woman's sensitivity to sexual side effects of HC use and also impact overall levels of desire.Further, some research finds that women who report more adverse childhood experiences are more likely to discontinue HCs due to diminished sexual function (Novick et al., 2022), which highlights that even childhood environments could confound HC use characteristics and sexual health.Health status, occupational characteristics, and stress are a few of the many other variables that should be considered as potential confounds in future research.
In addition to the variables mentioned above, relationships between sexual desire or partner attraction and each individual HC use and relationship characteristic tested in the current research likely possess their own set of unique unmeasured confounds.For example, a woman's original reason for using HCs could impact both type of product she used, as well as sexual health (e.g., as treatment for polycystic ovary syndrome vs. for pregnancy prevention).Timing of cycle resumption may be part of a complex causal web with sexual desire and sexual activity, wherein frequency of sex both influences-and is influenced by-the rate that ovarian function recovers and levels of sexual desire over time.Far from exhaustive, these examples just highlight the complexity of estimating causal relationships between HC use and sexual function, and the need to integrate well-established causal inference methods to move this research forward.
Future work that tracks sexual function over a longer time period prior to, during, and after HC use is needed to make definitive claims about if and for whom HC discontinuation impacts sexual and general health characteristics.If possible, the inclusion of a control group would aid in testing for causality, and the lack of a control group in the present study is another factor that precludes making casual inferences about the current data.The effects of attrition must also be considered.Only about a third of participants who completed Survey 1 continued on to complete Survey 2. We did not find that women who completed both surveys differed from those who did not on target outcomes at Survey 1 or demographic characteristics.However, these analyses do not rule out the possibility of non-response bias which should be considered when interpreting the results of the current research.
Further, in recent years, social media testimonies have increased women's concerns about beginning HC use (Kissling, 2016(Kissling, , 2014;;Le Guen et al., 2021) and made the possibility of negative HC-related side effects more salient to women using HCs.Accordingly, women might be choosing to discontinue their HC use with the expectation that they will experience increased sexual desire, and then reporting that they indeed feel more sexual desire as a result of the confirmation bias.Double-blind, placebo controlled clinical trials investigating the impact of HC use discontinuation on women's libido will be vital to address this possibility in future research.
Methodologically, the current study was limited in that we were only able to collect measures of women's sexual desire and partner attraction after women had already discontinued their HC use, and at two timepoints.In future work, it would be ideal to assess women's levels of sexual desire and partner attraction at regular intervals before HC use, during HC use, and after HC use discontinuation.This would allow researchers to determine if increases in these variables following HC use discontinuation were representative of a return to pre-HC use levels, or, if a different pattern of sexual desire and partner attraction might emerge.While the evidence suggests that the current results represent a return of libido following HC use discontinuation, it is impossible to state this without collecting data on women's sexual desire and partner attraction before and during HC use.
Additionally, we were only able to collect data on women at two timepoints following HC use discontinuation, which prevented us from making strong claims about the specific timing by which women's sexual desire increased following HC use discontinuation.Further, daily fluctuations in other variables, including but not limited to women's cycle phase, stress levels, mood, relationship conflict, and relationship satisfaction could have influenced women's ratings of their sexual desire and partner attraction at any given timepoint.Future research could mitigate this potential limitation by collecting more assessments of women's sexual desire and partner attraction over time, perhaps using intensive longitudinal data methods.
Finally, the current work relied upon women's self-reports of their sexual desire and partner attraction.Reliance on self-report measures introduces several potential sources of bias and measurement error.While we attempted to account for possible measurement error in our statistical models and controlled for several variables that may influence one's likelihood of inaccurately reporting sexual desire and partner attraction (e.g., age, goal), it cannot be ruled out that error or bias in selfreporting affected the present results.Future work would benefit from measuring women's sexual behavior, in addition to self-reported sexual desire and partner attraction, both to confirm the current findings and examine whether gains in libido post-HC use translate into increased sexual behavior.

Conclusions
Much research finds that HC use can have deleterious effects on sexual function and relationship satisfaction (Boozalis et al., 2016;Coelho and Barros, 2019;C.W. Wallwiener et al., 2010;M. Wallwiener et al., 2010).The present results extend this work by demonstrating that women experience increases in sexual desire that co-occur with resumption of the ovulatory cycles after discontinuing HCs use, with most women resuming cycling within three months.Further, this work provides insights into potential individual differences that moderate women's recovery timeline following cessation of HC use.Given limitations of the current data, including lack of measurement prior to and during HC use, the potential of unmeasured confounding, and lack of a control group, caution should be exercised when interpreting the results.Despite these limitations, the present findings lays the groundwork for future research to further investigate the effects HC use and discontinuation have on women's sexual health.

Fig. 1 .
Fig. 1.Example of natural cycles user interface.Fertility risk calculated based on body temperature changes across the cycle using a proprietary algorithm.

Fig. 2 .
Fig. 2. Diagram of sexual desire latent change score model.Shown here are standardized coefficients, as well as unstandardized coefficients (parentheses) to reflect fixed factor loadings across time per convention.Squares represent observed variables, and ovals represent latent factors.Solid arrows from latent factors to observed variables denote factor loadings, double-sided arrows denote a correlation, and arrows below observed variables denote error variances.Dashed arrows illustrate that relationships between covariates and each initial levels of sexual desire and change were tested.***p < 0.001.

Table 1
Characteristics of the sample.

Table 3
Summary of latent change score model results.
a Reflects analysis of levels of sexual desire/attraction at Survey 2.

Table 4
Frequency of any change in sexual desire and partner attraction.Shown here are means and standard deviations (parentheses) for each category and comparisons using one-way analyses of variance.HC = hormonal contraceptive; yrs.= years.*p < 0.05; **p < 0.01; ***p < 0.001.