Women's sexual experience during the menstrual cycle: identification of the sexual phase by noninvasive measurement of luteinizing hormone.

Women's sexuality is not limited to a specific period of estrus, as it is in almost all other mammals. Yet, ovarian hormones clearly modulate women's sexual behavior and even their subjective feelings of sexual desire and attractiveness. Previous research on variation in women's sexuality and emotional well being during the menstrual cycle has returned contradictory results. Many studies have identified a midcycle or periovulatory peak in sexuality (Adams, Gold, & Butt, 1978; Matteo & Rissman, 1984; Stanislaw & Rice, 1988; Udry & Morris, 1968), while other researchers report peaks right before or after menstruation (Bancroft, 1987) and still others report finding no identifiable peak (Meuwissen & Over, 1992). This lack of consensus is likely due to differences in methodology, including the methods for identifying and defining menstrual cycle phase, the aspects of sexuality measured, and the social and reproductive characteristics of the women studied.

Both the evolutionary advantage of coordinating sexual activity with fertility and the majority of existing literature point toward a midcycle increase in sexuality during the ovulatory phase. But, when measures of periovulatory hormones are coarse with error rates that create variation potentially greater than the phenomenon, an unequivocal pattern will be hard to detect. Moreover, the use of different methods to define and identify the hormonal phases of the ovarian cycle precludes direct comparisons across studies.

Menses and ovulation demarcate the hormonally distinct follicular and luteal phases. The most common method for estimating ovulation uses menses onset as the reference point. Participants are asked to make either retrospective reports (e.g., Bancroft, 1987; Warner & Bancroft, 1988) or prospective reports (e.g., Adams et al., 1978) relative to menses. Ovulation is estimated by counting backward 14 days from menses onset, based on the assumption that the luteal phase is relatively fixed in length (Udry & Morris, 1968). In fact, a functional luteal phase following an ovulatory luteinizing hormone (LH) surge varies from 4 to 19 days, introducing a 15-day error (Stern & McClintock, 1998). Counting forward from menses and assuming a 28-day menstrual cycle is even more imprecise, as the normal cycle in fact ranges from 24 to 35 days and the follicular phase varies from 3 to 19 days (Stern & McClintock, 1998).

Other common methods estimate the date of ovulation by measuring midcycle events such as a rise in basal body temperature (Stanislaw & Rice, 1988) or changes in cervical mucus characteristics, which indicate a shift from an estrogen to a progesterone predominant profile (reviewed in Stern & McClintock, 1995). The presence or absence of premenstrual syndrome (PMS) complaints have also been used as indirect indicators of hormonal profiles that may affect sexual behavior (Van Goozen, Wiegant, Endert, Helmond, & Vande Poll, 1997). The best measures are of plasma (Schreiner-Engel, Schiavi, Smith, & White, 1981; Van Goozen et al., 1997) or urinary (Hedricks, Piccinino, Udry, & Chimbira, 1987) hormone levels, but even these have yielded only rough estimates due to the timing of sample collection or interpretation of the data (reviewed in Stern & McClintock, 1995).

In the studies presented here, we changed the timing of a single urine sample collection to take into account the circadian and seasonal rhythms of the preovulatory LH surge. Traditionally, LH is measured in a urine sample collected in the early morning, upon rising. However, in most women the LH surge in urine begins after this and is not detectable until mid-morning. The LH surge measured by plasma usually begins in the early morning and reaches its peak in the evening (Baird, 1983; Baukloh, Fischer, Naether, & Bohnet, 1990; Baviera, Rigano, & Sfameni, 1988; Testart & Frydman, 1982; reviewed in Stern & McClintock, 1995). Moreover, there is a seasonal shift to the circadian pattern so that in the spring many women have LH surges that begin later in the day, around noon (Casper et al., 1988; Edwards, Steptoe, Fowler, & Baillie, 1980; Testart, Frydman, & Roger, 1982; reviewed in Stern & McClintock, 1995). Therefore, we hypothesized that the traditional early-morning collection time increases the risk of being a day or two late in detecting the LH surge, if not missing it altogether.

The traditional approach uses the peak of the LH surge as the predictor of the time of ovulation. However, it is the onset of the surge that is tightly coupled to ovulation, within 30 [+ or -] 2 hours of detection in urine, because the LH surge can have widely different shapes and durations and has been reported to peak from 2 days before ovulation to 4 days after ovulation (e.g., Baird, 1983; Baviera et al., 1988; extensive literature reviewed by Stern & McClintock, 1995). Thus, we tested the efficacy of using the first sample with detectable LH, rather than the sample with the highest level, to pinpoint timing of preovulatory LH surge and the day of ovulation. Our revised method was 95% accurate for detecting the onset within 12 hours as validated by six LH measures per day (see methods below and Stern & McClintock, 1995, for extended rationale and validation).

Another factor that could lead to disparate results is the measurement of different aspects of sexuality in different studies (e.g., frequency of intercourse, masturbation, orgasm, sexual arousal [Meuwissen & Over, 1992] fantasies, feelings of affection and desire, or lesbian sexual encounters [Matteo & Rissman, 1984]). We prospectively assessed a variety of psychosocial measures encompassing behavior and motivation including sexual activity, desire, fantasies, and feelings of loneliness in order to capture a more complete picture of sexuality. In addition, previous research has shown that female-initiated sexual behavior is more tightly coupled to the menstrual cycle than is male-initiated behavior (Adams et al., 1978). Thus, it is important to specifically ask about female initiation and record these variables separately.

Previous research has shown that knowing the hypothesis being tested and having personal beliefs about menstrual cycle correlates can create menstrual cycle patterns when none exist (Ruble, 1977). Therefore, it is essential to keep participants blind to the goals and hypotheses of this type of study to avoid findings colored by cultural or felt expectations about when sexual activity should occur, with whom, and the nature of the internal experience. We presented our studies in terms of the effect of the menstrual cycle on olfactory sensitivity to ensure that participants were blind to our goals and hypotheses.

The reproductive and social meanings of sexual behavior are essential contexts in which to study hormone-behavior relationships. In rhesus monkeys, social context interacts with hormonal mechanisms both when sexual activity serves to stabilize a consort pair (Wallen, Winston, Gaventa, Davis-Dasilva, & Collins, 1984) and when social groups negotiate the potentially "destabilizing" role of sexual behavior (Wallen, 2001). Indeed, in humans, "differences in the control and initiation of coitus in different groups and by marital status" are sociological phenomena (Udry & Morris, 1968, p. 593). Previous research in this area has shown the importance of collecting information about two aspects of social context: (a) who initiates the sexual behavior and (b) partnership status (Adams et al., 1978).

The reproductive context provides important additional information for questions about sexual activity and menstrual cycle phase. In research with humans, the variables of mind and intention cannot be ignored. Most women understand the reproductive consequences of sexual activity and can either target or avoid midcycle activity to achieve or avoid pregnancy. Therefore, the contradictory findings in previous research could arise from the diverse reproductive goals of different populations: for example, women training in natural family planning (Stanislaw & Rice, 1988), women on the birth control pill (Adams et al., 1978; Warner & Bancroft, 1988), lesbian couples (Matteo & Rissman, 1984), or a clinical population of women with premenstrual tension (Van Goozen et al., 1997).

Here we present two studies designed to improve upon these methodological shortcomings and inconsistencies. The participants were fertile women who were not taking the birth control pill, but who were trying to avoid pregnancy and were using barrier contraception. Thus, reproductive hormones were not being artificially regulated and in addition, detection of a peak in sexual activity around ovulation would not be an artifact of a woman trying to conceive. Both studies used a novel, noninvasive method for pinpointing the timing of the preovulatory LH surge with high accuracy and thus demarcating the follicular, ovulatory, and luteal phases, in addition to menses. In Study 1, this method was used to assess the frequency of sexual activity across the cycle and to compare the patterns of female and male initiation. Study 2 assessed sexual activity as well and focused on the variation in subjective sexual experience throughout the menstrual cycle. It also quantified the effects of having a stable sexual partner and general feelings of loneliness. The two studies enabled robust conclusions because they reproduced the basic findings using complementary yet independent methods.

A NONINVASIVE METHOD FOR PRECISELY IDENTIFYING PHASES OF THE MENSTRUAL CYCLE

Multiple Indicators of Menstrual Cycle Phase

The menstrual cycle, or the interval between two consecutive menses, is the biologically coherent unit of analysis for the human ovarian cycle (Spencer, Jacob, & McClintock, 2003). It is comprised of recruitment, selection, and growth of ovarian follicles, ovulation, transformation of the ruptured follicular tissue into a functional corpus luteum, and finally the death of the corpus luteum and the subsequent menses. The midcycle surge of luteinizing hormone (LH) is tightly coupled to ovulation (preceding it in urine by 30 [+ or -] 2 hours), and together they demarcate the transition from the estrogen-predominant follicular phase to the progesterone-dominant luteal phase. Ovulation and the associated dramatic change in hormonal profiles can be confirmed retrospectively by documenting high levels of progesterone produced by the functional corpus luteum.

This sequence of anatomical and hormonal events reliably changes the type of vaginal secretions and cervical mucus in all phases of the cycle and also increases basal body temperature (BBT) during the luteal phase. Although erroneously considered imprecise indicators of ovulation and menstrual cycles phase (reviewed by Stern & McClintock, 1998), changes in specific aspects of these characteristics are indeed correlated with LH surge onset when measured with our revised method described here. Thus, they can in fact be used as valid indicators of the LH surge onset, ovulation, and the associated cyclic changes in ovarian hormones, and they are particularly useful when hormonal data are unavailable or impractical to obtain.

Here we present a noninvasive method for precisely demarcating the hormonally distinct phases of the menstrual cycle. The precision of this method relies on revised methods for not only collecting urine, vaginal secretions, and temperature data, but also analyzing these samples and interpreting the results. Therefore, we asked participants to call the laboratory on the day they first experienced menstrual bleeding and again each day they had a positive LH urine test. They were then instructed on which days to collect three daily urine samples during the middle of their luteal phase. Throughout the entire menstrual cycle, they recorded daily the characteristics of their vaginal secretions and cervical mucus as well as their basal body temperature.

Preovulatory LH Surge in Urine

Participants were required to collect urine samples between 5 and 7 p.m. daily, beginning 1 week prior to the expected day of the preovulatory LH surge and ovulation (Baviera et al., 1988). We calculated the expected day of surge onset based on the regression line for the day of LH surge onset as a function of menstrual cycle length in our laboratory's database of menstrual cycle variables. These urine samples were analyzed by the participants with an enzyme immunoassay (Ovukit[TM], Conception Technology, San Diego, CA) in which the threshold for ovulatory levels of LH was greater than 35 milliInternational Units (mIU).

The noninvasive method for identifying the LH surge and thus pinpointing the day of ovulation we present here differs from traditional methods in two major ways. First, we shifted the sample time from a "first morning sample" collected upon arising to an early evening sample collected between 5 and 7 p.m. before dinner. We did this because the surge detectable in urine typically begins late in the morning or afternoon, after the traditional time for urine collection. We instructed participants to abstain from drinking fluids excessively between 3 p.m. and sample collection to avoid diluting their urine. Second, instead of using the peak of the preovulatory surge as is traditionally done, we used the onset of the surge as the predictor of the day of ovulation and the change in ovarian steroid production. We did this because the interval between the onset of the LH surge and ovulation is significantly less variable than the interval between the peak of the surge and ovulation (reviewed in Stern & McClintock, 1995).

These two changes in timing of sample collection and data interpretation made our method for pinpointing the timing of the LH surge and thereby ovulation significantly more accurate than the traditional method (94% correct vs. 22% correct; p < .001, [chi square] = 35.3, N = 38 cycles, see Figure 1). In a validation study, we created a referent by collecting urine samples six times throughout the day beginning at least 1 day before the preovulatory LH surge until its end. Absolute LH levels in urine were measured with a radioimmunoassay. We selected a subset of the samples, assayed them with a threshold-based enzyme immunoassay, and interpreted the results following the revised protocol described above. We collected another subset and interpreted it following the traditional protocol. A third subset used samples collected at the traditional time, but we interpreted them using the first day of detectable LH rather than the day of its highest level. The accuracy of each protocol was assessed by comparing it to the referent established by the six samples collected each day for several days. In sum, our revised method enables precise identification of the onset of the preovulatory LH surge (onset time within 12 hours with 95% accuracy; Stern & McClintock, 1995). We have used this method successfully for over a decade and have excellent subject compliance (> 95%).

[FIGURE 1 OMITTED]

Estimating the Onset of Preovulatory LH Surge

Vaginal secretions and cervical mucus. Participants recorded in a daily journal three characteristics of their vaginal secretions: consistency, color, and amount. Each of these characteristics had a significant, albeit variable, relationship to the day of ovulation (Stern, 1992). Therefore, we developed an algorithm using all three characteristics that was more accurate than one based on any one characteristic alone. The change in consistency of vaginal secretions was the most accurate for pinpointing the day of ovulation (Stem, 1992). Ovulation occurred most often on the day before the consistency of vaginal secretions changed from wet, slippery, or spinnbarkheit (clear and pliable strands) to tacky, sticky, or dry. If the consistency measure yielded several potential days of ovulation, we used the day that was also indicated by a change in the color characteristic. Ovulation typically occurs on the first day color changes from glassy and/or clear (i.e., estrogenized cervical mucus) to yellow and/or white (i.e. progesterone-influenced and nonestrogenized vaginal cells). Finally, if the consistency and color characteristics did not agree, we used the day indicated by a change in the amount of secretions. Ovulation typically occurs a day before a decrease in amount.

Basal body temperature. Participants were instructed to take their basal body temperature (BBT) in the morning when they first awoke before undertaking any other activity and as close to the same time each day as possible. They took their temperature orally with a digital thermometer and recorded it in the daily logbook.

To determine the day of ovulation, BBT profiles were charted using a 3-day moving average. Each of two raters independently sorted the profiles into three types: classic biphasic profile, midcycle peak profile, and irregular profile. A biphasic curve profile (Martinez et al., 1992) is the most typical BBT profile (76%). It has a clearly visible biphasic curve including a low plateau following menstruation, a midcycle increase in temperature occurring over 2 subsequent days, and then a higher temperature plateau indicating a functional corpus luteum. A midcycle peak profile (11%) has a clearly visible midcycle peak consisting of a low plateau following menstruation, a steep 2-day increase in temperature, and then a steep 2-day drop in temperature followed by a low temperature plateau. A profile was deemed irregular (13%) if it (a) undulated over and under the cycle average line so that there were two or more peaks, (b) was flat across the whole profile, (c) consisted of a double rise (two steep rises in temperature, with the intervening temperatures right on the cycle average line), or (d) if a significant amount of BBT data were unusable (due to fever or noncompliance).

For the classic biphasic profile, the day of the LH surge was estimated using a method found to be the most accurate of those derived from basal body temperatures (Stem, 1992). Specifically, we graphed a 3-day moving average of temperatures. The day of the preovulatory LH surge occurs 3 days before the smoothed curve rises above the average cycle temperature (Stem, 1992). Prior studies have not estimated BBT profiles that are not biphasic; however, our revised method indicated that the day of LH surge is the day halfway between the two peaks.

Interpreting multiple measures. The cervical mucus and BBT criteria described above were used to generate a combined estimate of the day of the preovulatory LH surge. Specifically, we considered several sources of information including (a) the vaginal secretions and BBT estimates described above, (b) confidence in the particular vaginal secretion and/or BBT estimate (e.g., BBT patterns that were irregular were not assigned a high confidence rating), and (c) the relative difference in days between the individual cervical mucus and BBT estimates, in which case an algorithm was used to provide a combined estimate of the preovulatory LH surge.

Two trained judges independently provided a combined estimate of the preovulatory LH surge date. If the two judges' estimated dates were the same, that date was accepted as the estimated preovulatory LH surge date for that cycle. If the judges' estimates of the preovulatory LH surge date were not the same, the judges met, reviewed the BBT and cervical mucus data as well as any other available data (including urine concentration of pregnanediol glucuronide and the traditional assumption of a 14-day luteal phase) to come to a consensus regarding the final preovulatory LH surge date for that cycle. This procedure was done for cycles with and without an established urine-derived preovulatory LH surge date to determine accuracy of the LH surge estimates derived from vaginal secretions and BBT. When both raters' final estimates were combined using this method, 88% of predictions from the 48 cycles with known urine LH surges were within 2 days of the urine-derived LH surge onset day, and 94% of the 48 cycles were within 3 days of the urine-derived day. This is significantly higher than previously thought and can establish a periovulatory phase with high confidence.

Confirming Ovulation

Participants collected urine on days 5, 7, and 9 after the onset of the preovulatory LH surge to determine levels of pregnanediol-3-glucuronide. These were measured either by an enzyme immunoassay (ProgestURINE kit; Monoclonol Antibodies, Inc.) or radioimmunoassay. A functional corpus luteum produces levels above 1.5 [micro]g/ml or 4 ng/ml respectively.

On rare occasions, women can have a surge of preovulatory LH but not ovulate or form a fully functional corpus luteum (reviewed in Stern & McClintock, 1995). These are anovulatory cycles with atypical hormonal profiles. Therefore, when studying the association between hormones of an ovulatory menstrual cycle and any variable, it is essential to confirm the development of a functional corpus luteum from the ovulatory follicle(s). A rise in urinary pregnanediol-3-alpha-glucuronide above the assay-specific threshold indicates ovulation has occurred.

We fully recognize that, very rarely, a corpus luteum can form even when the ovulatory process is abnormal. Moreover, truly documenting ovulation itself requires visualization through laproscopy or conception, neither of which is feasible in noninvasive studies of the menstrual cycle. Thus, conservatively, what we term "the day of ovulation" should be termed "the day when the follicle became a corpus luteum." With this caveat we chose the former for the sake of brevity.

Menstrual Cycle Days

To compare hormonally similar days across participants, each participant's cycle was justified three ways relative to her preovulatory LH surge and to menses onset. First, the days were numbered relative to the LH surge (LH day 0), running both backwards (e.g., LH day -1 to LH day -14) and forwards (e.g., LH day +1 to LH day +14). Second, menses days were defined as cycle day +1, the first day of menstruation, cycle day +2, the next day, and so on until the day preceding the subject's next menstruation. Finally, premenstrual day - 1 was defined as the day before menses, running backwards until premenstrual day -4, 4 days before menses.

Menstrual Phase Definitions

Early menses is the first 3 days of menstruation (cycle days 1-3) inclusive. Menses onset was defined as the first day during menstruation that any pink or brown mucus, blood spotting, or blood was reported in vaginal secretions. Basal body temperature is typically relatively low at this time. Early menses is characterized by a hormonal profile of low levels of progesterone and estrogens and low levels of LH and follicle-stimulating hormone (FSH).

Late menses consists of cycle day 4 until the end of menses, which is defined as the first day without any pink or brown mucus, blood spotting, or blood reported in vaginal secretions. Basal body temperature is also relatively low at this time. Late menses is typically characterized by a hormonal profile of low levels of progesterone and estrogens and low levels of LH and FSH.

The follicular phase is comprised of the first day without menses through the day prior to the day of preovulatory LH surge onset (LH day--1). At this time, vaginal secretions are often opaque (white or yellow), thick, and tacky. Basal body temperature is typically low at this time. FSH causes the growth of an ovarian follicle so that the follicular phase is characterized by a hormonal profile of rising blood levels of estrogen from the developing follicle.

The ovulatory phase is the day of the LH surge onset and the two subsequent days (LH day 0 to +2). The length of this phase is fixed at 3 days because it is well established in the literature that ovulation occurs 30 + 2 hours after the onset of the LH surge in a urine sample (Baviera et al., 1988; Stern & McClintock, 1995). The LH surge and subsequent ovulation mark the anatomical and hormonal transition from predominance of estrogen produced by the developing follicle to the predominance of progesterone produced by the corpus luteum. During this phase, vaginal secretions are more profuse and are typically thin, watery, and clear. The specific type of stretchable, thread-like mucus, spinnbarkheit, occurs at this time. Basal body temperature rises at this end of this phase. FSH, as well as LH, peaks at this time. Circulating levels of testosterone are often greatest during this phase as well.

The early luteal phase includes LH day + 3 through premenses day--4. Vaginal secretions may return to dry or thick at this time. Basal body temperature is typically from 0.3 to 1.0 degree higher relative to the follicular phase during this time. The early luteal phase is dominated by progesterone.

The late luteal phase consists of the 3 days before menses onset (premenses days--3 to--1). During this phase the corpus luteum is regressing, accompanied by falling levels of progesterone. Similar to the early luteal phase, vaginal secretions may be dry and thick and basal body temperature is relatively high, falling prior to menses onset.

EXPERIMENTAL PROTOCOL

We used the following protocol for both Study 1 and Study 2. Specific details unique to each study, such as sample size and additional measures, are presented in the study methods sections.

Participants

Participants in both studies were healthy women between the ages of 18 and 35 years who were currently having typical spontaneous menstrual cycles (24-34 days long). The women were recruited through posters and fliers on the campus of a major midwestern university, and through advertisements in community newspapers. They were blind to the purpose of this study (sexual motivation) since the advertisements stated that the study was about olfactory acuity and the menstrual cycle.

The women accepted into the study had a history of regular menstrual cycles, were not using birth control pills or an IUD, and had never been pregnant. In addition, they reported they were not taking prescription or nonprescription drugs on a regular basis and were not experiencing psychiatric disorders. Participants were measured and weighed to ensure that they were within 30% of normal height-to-weight ratio. We collected data regarding exercise types and duration on a daily basis. Participants also agreed to use only barrier contraception and were planning not to become pregnant during the study period. All participants participated on a voluntary basis and were paid a sum approved by the Institutional Review Board at The University of Chicago.

Recruitment and Blinding Procedures

After a structured telephone screening, each participant came to the laboratory for training in tracking their menstrual cycles, odorant testing, and completing daily logs of olfactory and psychosocial information. The explicit purpose of the study was presented in detail as a study of the effects of the menstrual cycle on changes in olfactory sensitivity during the context of everyday life (e.g., Doty, Snyder, Huggins, & Lowry, 1981). They were told that these odors would be mild, at or below the level of conscious detection. During the consent process, they were given a list of 70 natural compounds that they might receive. In fact, all participants were given cotton pads without a specific odorant, although the pads themselves did have a mild odor.

Each participant was taught to fill out a prospective daily logbook in which she recorded the daily menstrual cycle data described above. Throughout the study, she also recorded psychosocial and behavioral data about her activities that day. To ensure compliance and maintain the blinding procedure, we asked participants to come to the laboratory twice per week. During these visits, the experimenter ensured that they were collecting data in the appropriate way and gave them additional "odor" pads. At the end of the study, participants completed an out-take interview and gave additional information on sexual orientation and current romantic partner status.

Measurement of Sexual Activity

Participants were asked to record in their daily logbook occurrence of a wide variety of physical activities, physical symptoms, and illnesses. In the midst of this extensive form was an item to check if they had engaged in sexual activity that day. Sexual activity was defined as including both activity with another person and masturbating alone. When activity was with a partner, participants also indicated who initiated the activity: oneself, one's partner, or both. No subject detected an experimental focus on sexual activity or motivation.

STUDY 1

In Study 1, we asked whether precisely defining the phases of the menstrual cycle with reference to the onset of the preovulatory LH surge would reveal a periovulatory or premenstrual increase in sexual activity, particularly when initiated by the woman.

Methods

To answer this question, 19 women between the ages of 20 and 32 years provided daily sex and menstrual cycle data for up to five menstrual cycles, with a total of 57 cycles contributed from the 19 women. A good estimate of the change in each woman's behavior across menstrual cycle phase was obtained by collecting data from multiple cycles and summarized for each woman as the percent of cycles with sexual activity on a given day. This also addressed the issue of nonindependence arising from multiple cycles from the same women. Thirteen study participants identified themselves as single, one lived with her partner, and five were married. All women reported sexual activity during at least one phase of the menstrual cycle.

Results

Rate of sexual activity showed a clear pattern relative to the onset of the preovulatory LH surge. It rose during the 4 days prior to the LH surge onset and remained high a few days afterward (see shaded area in Figure 2). To present the absolute rates of sexual activity, the proportion of a woman's cycles with sexual activity on a given day was plotted with a three-day moving average ([+ or -] SEM; sexual activity included all occasions of activity regardless of who initiated it).

[FIGURE 2 OMITTED]

Sexual activity varied significantly across the phases of the menstrual cycle, F (5,90) = 2.53, p [less than or equal to] .03, repeated measures analysis of variance. (To adjust for large individual differences in average rate of sexual activity, this analysis was based on z scores for each phase, normalized to each woman's mean.) Sexual activity was highest during the follicular phase of the menstrual cycle (z = 0.47 [+ or -] 0.19) and lowest during the first 3 days of menstruation (z = -0.28 [+ or -] 0.14). Sexual activity in the follicular phase was significantly higher than in each of the other phases: early menses (-0.28 [+ or -] 0.14, p [less than or equal to] 0.0005), late menses (-0.09 [+ or -] 0.17,p [less than or equal to] 0.009), ovulatory phase (0.04 [+ or -] 0.15,p [less than or equal to] 0.04), early luteal phase (0.01 [+ or -] 0.11, p [less than or equal to] 0.003), and late luteal phase (-0.15 [+ or -] 0.11, p [less than or equal to] 0.004; Fisher PLSD Tests). Female-initiated sexual activity was associated with menstrual cycle phase, peaking during a 3-day period ending on the day of LH surge onset and showing a secondary rise in the luteal phase (see Figure 3). In contrast, the distribution of male-initiated or mutually initiated sexual activity did not vary significantly with respect to the menstrual cycle. In sum, the days of highest total sexual activity centered around the LH surge, or, when focusing just on female-initiated activity, on the day of LH surge onset and the 2 preceding days. In both measures, maximum sexual activity straddles the late follicular and ovulatory phase of the cycle. Thus we propose referring to this behaviorally defined phase of the cycle as the "sexual phase" (shaded areas in Figures 2 and 3), rather than using follicular and ovulatory phases, which are terms based on ovarian events and are overly inclusive.

[FIGURE 3 OMITTED]

STUDY 2

Study 2 was designed to replicate and extend the characterization of the sexual phase of the menstrual cycle. We quantified sexual motivation in terms of women's sexual desire and sexual fantasies. We expected these subjective experiences to have a tight relationship with hormonal state, because they are less constrained by circumstance than is sexual behavior. The women in Study 1 had been single or married. Because having a regular sexual partner is an important contextual variable likely to affect sexual activity and motivation, this study recruited a sample of women large enough to assess the effects of partnership on women's sexual experience. We also measured feelings of loneliness, as it is typically higher in women without a stable relationship and yet might be ameliorated by sexual activity.

Methods

Measures and procedure. Forty-six women each contributed data from one menstrual cycle. Menstrual cycle phases were determined as described above in the section "A Noninvasive Method for Precisely Identifying Phases of the Menstrual Cycle." In addition to providing daily menstrual cycle and sexual activity data, participants also answered daily log book questions to quantify the following as continuous psychosocial variables: strength of sexual desire, number of sexual fantasies, and intensity of loneliness. These logbook entries were completed each evening and contained a multitude of items concerning physical activity, symptoms, illnesses, and psychosocial variables. No subject detected our focus on sexuality.

Sexual activity. Sexual activity was defined during the training session as any sexual activity either with a partner or alone, and thus included autoerotic and other sexual activities in addition to sexual intercourse. Participants were also asked to indicate whether the sexual activity was initiated by oneself, one's partner, or equally by both, and to indicate what if any birth control methods were used.

Sexual desire. To assess strength of desire for sexual intimacy, participants were asked to indicate on a visual analog scale "the degree (frequency, intensity or duration) to which you felt desire today for sexual intimacy with other people (e.g. your partner, friends, strangers, etc.)". The poles were labeled not at all and very much. The strength of desire was quantified by measuring to the nearest millimeter the length of the line and the point on that line marked by the participant, and calculating the proportion to create a scale between 0 and 1.0.

Sexual fantasies. The sexual fantasy variable was assessed using a yes/no question in which participants were asked whether they recalled "having any fantasies/daydreams today which were of a sexual or romantic nature" and, if so, to estimate the number. If a participant gave a range for the number of fantasies (e.g., 4-6) the mean was used. We found large individual differences in numbers of fantasies reported, which were potentially greater than the effect of menstrual cycle hormones. To standardize for these individual differences, we based data on a z score calculated for each woman's daily average.

Loneliness. We assessed psychosocial loneliness with a single item in the logbook, which asked participants to check a box if they had felt lonely that day.

Partnership status. Whether a woman was considered to have a sexual partner during the cycle was determined using two sets of criteria. Women considered "partnered" for a cycle either (a) slept in the same bed as another person identified as "boyfriend, girlfriend or husband" at least 15% of the nights of the cycle, and review of sexual activity data indicated an increase of activity on those days; or (b) did not meet criteria a but review of sexual activity data and retrospective self-reported partnership status clearly indicated that they had a regular sexual partner during that cycle.

Results

Daily pattern. The increase in sexual activity also appeared prior to the LH surge onset in this sample (see Figure 4: a 3-day moving average [+ or -] SEM as in Figure 1; sexual activity included all occasions of activity regardless of who initiated it). Because each subject was studied for only one cycle, z scores were used to account for individual differences in average activity. Again, there was a gradual increase in reported sexual activity from low levels during and after menstruation leading to a sharp rise during the late follicular phase. This was followed by a gradual decrease in sexual activity over the next several days. The sexual phase of maximal activity was the 3-day period centered on the LH surge. Sexual activity initiated by women in this sample showed a similar pattern to that seen in Study 1, declining after the detection of the LH surge. The distribution of male initiated activity did not manifest a similar pattern with respect to the menstrual cycle (see Figure 5). In addition, there was a secondary rise prior to menses.

[FIGURES 4&5 OMITTED]

Women's subjective experience of sexual motivation increased midcycle (see Figure 6). Strength of desire for sexual intimacy gradually increased from relatively low levels following menses to a peak near the onset day of the preovulatory LH surge. Both measures rose again prior to menses. In contrast, felt loneliness was lowest at this time. There is a small negative relationship between loneliness and sexual activity (loneliness was never reported on days with sexual activity vs. 7.9% of days without, [chi square] = 4.0, p [less than or equal to] 0.04). The categorical measure of simply having had any sexual fantasies or not did not vary systematically during the cycle, indicating that the continuous scales (number of sexual fantasies and strength of desire) are more sensitive measures of subjective sexual experience in this population.

[FIGURE 6 OMITTED]

Cycle phase and partner status. To test the hypothesis that the effect of menstrual cycle phase would be seen in women with regular sexual partners, we conducted repeated measures ANOVAs using cycle phase and partner status as independent variables on the four dependent variables of interest: sexual activity, desire, number of fantasies, and loneliness.

Sexual activity peaked during the ovulatory phase of the menstrual cycle, which began with LH surge onset and included the next two days during which ovulation occurred (see Figure 7; F [5, 195] = 3.61, p [less than or equal to] .05), and was significantly higher than during early and late menses and early and late luteal phases (Fisher's PLSD, p [less than or equal to] 0.002, p [less than or equal to] 0.01, p [less than or equal to] 0.04, p [less than or equal to] 0.01 respectively). This pattern was more pronounced in women with regular sexual partners, F(5,195) = 2.37, p [less than or equal to] .04, who also had more sexual activity overall, F(1, 39) = 16.46, p [less than or equal to].0002.

[FIGURE 7 OMITTED]

Self-reported desire for sexual intimacy also varied significantly during the menstrual cycle (See Figure 5; F [5, 175] = 4.33, p [less than or equal to] .001). A rise was first detectable between early menses and the follicular phase (Fisher's PLSD, p [less than or equal to] .03) and was strongest in the ovulation phase, in comparison with late menses and the follicular, early luteal, and late luteal phases (Fisher's PLSD, p [less than or equal to] .01, p [less than or equal to] .04, p [less than or equal to] .07, and p [less than or equal to] .002 respectively). Partnered women only tended to report stronger desire for sexual intimacy than did those without regular partners, F(1, 35) = 3.22, p [less than or equal to] .08. Moreover, the absence of a statistically significant interaction between cycle phase and partnership status indicated that the effect of menstrual cycle on sexual desire was independent of having a regular sexual partner, F(5, 175) = 1.59, p [less than or equal to] .16.

For number of sexual fantasies, however, there was a significant interaction, between menstrual cycle phase and partnership status (See Figure 7; F[5, 185] = 2.28, p [less than or equal to] .04). Only women with partners reported more sexual fantasies during the ovulatory phase. Overall, however, women with and without partners reported similar numbers of sexual fantasies, F(1, 37) = 1.43, p [less than or equal to] .24, and there was no main effect of cycle phase, F(5, 185) = .94, p [less than or equal to] .54.

Women without partners reported feeling lonely more often than did women with partners F[1, 39] = 5.14, p [less than or equal to] .03; menstrual cycle phase, F[1, 39] = 0.23, ns, interaction F = 0.45, ns), except during the ovulatory phase when the nonpartnered women's loneliness was at its nadir and no longer significantly higher than partnered women's (t = 1.2, ns). This ovulatory nadir in loneliness was not present in women with regular sexual partners.

DISCUSSION

Our method of pinpointing the onset of the preovulatory LH surge enabled detection of a rise in sexual activity preceding the LH surge. The finding is robust given that it was seen in two independent samples of women who did not wish to become pregnant at the time of the study, were using less-reliable barrier contraception, and were unaware of the hypothesis being tested. A peak in sexual motivation surrounding the LH surge was detected in women with and without regular sexual partners, demonstrating that it is also robust across different social conditions.

The precision of our hormonal measure allows us to conclude that sexual activity and motivation is significantly elevated for 3 days prior to the LH surge and 2 days afterward, which we have termed the sexual phase of the menstrual cycle (see shaded areas in the figures). The timing of this phase was based on a composite of all measures in these studies. Sexual activity at this time is optimal for conception, because sperm can live in the reproductive tract for several days; ovulation follows the onset of the preovulatory surge well within this timeframe (30 [+ or -] 2 hours), and intercourse after ovulation with concomitant cervical stimulation may enhance function of the corpus luteum. That the increase in sexual motivation precedes the LH surge and ovulation is particularly adaptive because human sperm live 3 to 5 days in the reproductive track while the egg is viable for only 24 hours. Thus, the rise in a woman's sexual motivation prior to the LH surge increases the chance that viable sperm will be present in the oviduct when she ovulates.

The hormonal milieu surrounding the onset of the LH surge clearly increases women's sexual motivation. Women were more likely to initiate sexual activity, had stronger sexual desire whether or not they had steady sexual partners, and had more sexual fantasies. There was no evidence in this study for a periovulatory increase in sexual attractiveness as occurs in other primates (Domb & Pagel, 2001), because there was no evidence for an increase in male-initiated activity.

During the 6-day interval bracketing the onset of the LH surge, estrogen has been rising for several days, undoubtedly changing receptor levels in key brain areas in humans as it does in many mammals. Estrogen may indeed be both necessary and sufficient to sustain active sexual behavior as it does in rhesus monkeys (Zehr, Tannenbaum, Jones, & Wallen, 2000). The onset of the LH surge is coincident with a rise in progesterone and a fall in estradiol levels, as well as preceded by a more prolonged rise in 5-alpha hydroxyprogesterone (Caruso et al., 1989; Dawood & Saxena, 1976; Guerrero et al., 1976; Testart & Frydman, 1982; Uehara et al., 1985). Thus, dynamic interaction between estrogens and progesterones may underly women's sexual behavior, as it does in the rhesus monkey and Norway rat (Walker, Gordon, & Wilson, 1983). Indeed, progesterone, 5-alpha hydroxyprogesterone, and allopregnanolone are all selectively concentrated in the hypothalamus and amygdala and vary with ovarian steroid production (Bixo, Andersson, Winbald, Purdy, & Backstrom, 1997). This hypothesis is supported by the secondary rise in sexual activity in the late luteal phase (see Figures 4 and 6), when estrogen has been high and progesterone is falling from inhibitory levels into an intermediate range. A similar pattern is found in pregnant rhesus monkeys when they have a brief estrus period alter progesterone from the corpus luteum of pregnancy is falling and before placental progesterone has risen to inhibitory levels (Wilson, Gordon, & Collins, 1982).

There is also a transitory increase in testosterone during the periovulatory phase (Caruso et al., 1989; Dawood & Saxena, 1976; Guerrero et al., 1976; Treloar, Boynton, Behn, & Brown, 1967; Uehara et al., 1985). Therefore, our data are consistent with Bancroft and Sherwin's hypothesis that women's sexual motivation is enhanced by the concurrent action of testosterone and estradiol (Bancroft, Sherwin, Alexander, Davidson, & Walker, 1991a, 1991b; Sherwin, Gelfand, & Brender, 1985). Whether this is mediated by testosterone freeing estradiol from their common sex hormone binding globulin (K. Wallen, personal communication, September 1995) awaits experimental manipulation. In addition, other neuroendocrine changes occurring at this time must also be evaluated, such as potential contributors to women's sexual experience: alpha melanocyte stimulating hormone, LHRH, inhibin, prolactin, oxytocin, and many others (Argiolas, 1999; Mauri et al., 1990; Shukovski, Healty, & Findlay, 1989; Tsonis, Messinis, Templeton, McNeilly, & Baird, 1988; Werawatgoompa et al., 1981).

Our data confirm the importance of assessing the social context of sexual behavior. Having a regular sexual partner increased the amount of sexual activity overall as well as the desire for sexual intimacy. Nonetheless, the menstrual cycle effect on sexual activity was equally strong in women with and without regular sexual partners. Unpartnered women had the same number of sexual fantasies as those with partners (except around ovulation) and, perhaps consequently, were more likely to feel lonely. This hypothesis is supported by the finding that unpartnered women felt less lonely during the periovulatory phase when they were more likely to have sexual activity. Future studies are needed to determine whether this periovulatory drop in loneliness is indeed a consequence of increased sexual activity or a direct hormonal effect of the ovarian axis. In addition, future studies could explore specific aspects of sexual motivation in more detail than we were able to here, due to our commitment to keep participants free of potentially biasing information regarding study goals and hypotheses.

Using this novel protocol to measure and interpret urinary LH, together with basal body temperature, cervical mucus, and vaginal secretions, enabled precise identification of hormonally distinct phases of the menstrual cycle as well as the sexual phase of the cycle that spans the late follicular and early ovulatory phases. This method can now be successfully applied to identifying other psychological and social correlates of the menstrual cycle and definitions of menstrual cycle phases. Equally important will be its application to assessing the effects of social, environmental, and pharmacological agents on ovarian function, where it will be possible to distinguish effects on the length of follicular development, timing of the preovulatory LH surge, and life span of the corpus luteum.

REFERENCES

Adams, D. B., Gold, A. R., & Burr, A. D. (1978). Rise in female-initiated sexual activity at ovulation and its suppression by oral contraceptives. New England Journal of Medicine, 299, 1145-1150.

Argiolas, A. (1999). Neuropeptides and sexual behaviour. Neuroscience & Biobehavioral Reviews, 23, 1127-1142.

Baird, D. T. (1983). Prediction of ovulation: Biophysical, physiological and biochemical coordinates (Vol. 3). New York: John Wiley & Sons.

Bancroft, J. (1987). Hormones, sexuality and fertility in women. Journal of Zoology, 213, 445454.

Bancroft, J., Sherwin, B. B., Alexander, G. M., Davidson, D. W., & Walker, A. (1991 a). Oral contraceptives, androgens, and the sexuality of women: II. The role of androgens. Archives of Sexual Behavior, 20, 121-135.

Bancroft, J., Sherwin, B. B., Alexander, G. M., Davidson, D. W., & Walker, A. (1991b). Oral contraceptives, androgens, and the sexuality of young women: I. A comparison of sexual experience, sexual attitudes, an gender role in oral contraceptive users and nonusers. Archives of Sexual Behavior, 20, 105-120.

Baukloh, V., Fischer, R., Naether, O., & Bohnet, H. G. (1990). Patterns of serum-luteinizing hormone surges in stimulated cycles in relation to injections of human chorionic gonadotropin. Fertility and Sterility, 53, 69-75.

Baviera, G., Rigano, A., & Sfameni, E (1988). 4-Hour urinary and serum immunofluorimetric assay for luteinizing hormone to detect onset of preovulatory LH surge. Acta Europaea Fertilitatis, 19, 25-28.

Bixo, M., Andersson, A., Winbald, B., Purdy, R. H., & Backstrom, T. (1997). Progesterone, 5alpha-pregnane-3,20-dione and 3alpha-hydroxy5alpha-pregnane-20-one in specific regins of the human female brain in different endrocrine states. Brain Research, 764, 173-178.

Caruso, A., Lanzone, A., Fulghesu, A. M., Apa, R., Guida, C., & Mancuso, S. (1989). Importance of echographic and endocrine monitoring for the assessment of ovulation by follicle stimulating hormone in polycystic ovarian disease. Journal of Gynnaecology & Obstetrics, 28, 163-169.

Casper, R. E, Erskine, H. J., Armstrong, D. T, Brown, S. E., Daniel, S. A., Graves, G. R., et al. (1988). In vitro fetilization: Diurnal and seasonal variation in luteinizing hormone surge onset and pregnancy rates. Fertility and Sterility, 49, 644-647.

Dawood, M. Y., & Saxena, B. B. (1976). Plasma testosterone and dihydrotestosterone in ovulatory and anovulatory cycles. American Journal of Obstetrics, 126, 430-435.

Domb, L. G., & Pagel, M. (2001). Sexual swellings advertise female quality in wild baboons. Nature, 410, 204-206.

Doty, R. L., Snyder, P. J., Huggins, G. R., & Lowry, L. D. (1981). Endocrine, cardiovascular, and psychological correlates of olfactory sensitivity changes during the human menstrual cycle. Journal of Comparative & Physiological Psychology, 95, 45-60.

Edwards, R. G., Steptoe, P. C., Fowler, R. E., & Balllie, J. (1980). Observations on preovulatory human ovarian follicles and their aspirates. Journal of Obstetrics and Gynaecology, 87, 769-779.

Guerrero, R., Aso, T., Brenner, P. E, Cekan, Z., Landren, B. M., Hagenfeld, K., et al. (1976). Studies on the pattern of circulating steroids in the normal menstrual cycle. I. Simultaneous assays of progesterone, pregnenolone, dehydroepiandrosterone, testosterone, dihydrotestosterone, androstenedione, oestadiol and oestroue. Acta Endocrinologica (Copneh), 81, 133-149.

Hedricks, C., Piccinino, L., Udry, J., & Chimbira, T. (1987). Peak coital rate coincides with onset of luteinizing hormone surge. Fertility and Sterility, 48, 234-238.

Martinez, A. R., van Hooff, M. H., Schoute, E., van der Meer, M., Broekmans, E J., & Hompes, P. G. (1992). The reliability, acceptability and applications of basal body temperature (BBT) records in the diagnosis and treatment of infertility. European Journal of Obstretrics & Gynecology and Reproductive Biology, 47, 121-127.

Matteo, S., & Rissman, E. E (1984). Increased sexual activity during the midcycle portion of the human menstrual cycle. Hormones and Behavior, 18, 249-255.

Mauri, A., Martellotta, M. C., Melis, M. R., Caminiti, E, Serri, E, & Fratta, W. (1990). Plasma alpha-melanocyte-stimulating hormone during the menstrual cycle in women. Hormone Research, 34, 66-70.

Meuwissen, I., & Over, R. (1992). Sexual arousal across phases of the human menstrual-cycle. Archives of Sexual Behavior, 21, 101-119.

Ruble, D. N. (1977). Premenstrual symptons: A reinterpretation. Science, 197, 291-292.

Schreiner-Engel, P., Schiavi, R. C., Smith, H., & White, D. (1981). Sexual arousability and the mentrual cycle. Psychosomatic Medicine, 43, 199-214.

Sherwin, B. B., Gelfand, M. M., & Brender, W. (1985). Androgen enhances sexual motivation in females: A prospective, crossover study of sex steriod administration in the surgical menopause. Psychosomatic Medicine, 47, 339-351.

Shukovski, L., Healty, D. L., & Findlay, J. K. (1989). Circulating immunoreactive oxytocin during the human menstrual cycle comes from the pituitary and is estradiol dependent. Journal of Clinical Endocrinology and Metabolism, 68, 455-460.

Spencer, N., Jacob, S., & McClintock, M. K. (2003). Stable variability: Homeostasis of periodicity in the human menstrual cycle. Unpublished manuscript, University of Chicago.

Stanislaw, H., & Rice, E J. (1988). Correlation between sexual desire and menstrual-cycle characteristics. Archives of Sexual Behavior, 17, 499-508.

Stern, K. (1992). Ovulation in women: Methods of detection, behavioral correlates, and pheromonal regulation. Chicago: University of Chicago.

Stem, K., & McClintock, M. K. (1995). Biological rhythms and individual variation: Accurate measurement of menstrual cycle phase and the pre-ovulatory LH surge. In M. F. Jensvold, U. Halbreich, & J. Hamilton (Eds.), Psychopharmacology of women (pp. 393-4 13). Washington, DC: American Psychiatry Press.

Stern, K., & McClintock, M. K. (1998). Regulation of ovulation by human pheromones. Nature, 392, 177-179.

Testart, J., & Frydman, R. (1982). Minimum time lapse between luteinizing hormone surge or human chorionic gonadotropin administration and follicular rupture. Fertility and Sterility, 37, 50-53.

Testart, J., Frydman, R., & Roger, M. (1982). Seasonal influence of diurnal rhythms in the onset of the plasma luteinizing hormone surge in women. Journal of Clinical Endocrinology and Metabolism, 55, 374-377.

Treloar, A. E., Boynton, R. E., Behn, D. G., & Brown, B. W. (1967). Variation of the human menstrual cycle through reproductive life. International Journal of Fertility, 12, 77-126.

Tsonis, C. G., Messinis, I. E., Templeton, A. A., McNeilly, A. S., & Baird, D. T. (1988). Gonadotropic stimulation of inhibition secretion by the human ovary during the follicular and early luteal phase of the cycle. Journal of Clinical Endocrinology and Metabolism, 66(5), 915-921.

Udry, J. R., & Morris, N. M. (1968). Distribution of coitus in menstrual cycle. Nature, 220, 593.

Uehara, S., Naganuma, T., Chida, S., Yohkaichiya, T., Tsuiki, A., Kyohno, K., et al. (1985). The evaluation of serum estradiol, progesterone, testoterone concentrations and urinary estrogen excretion level for monitoring indices of follicular maturation. Nippon Sanka Fujinka Gakkai Zasshi, 37, 1849-1858.

Van Goozen, S. H. M., Wiegant, V. M., Endert, E., Helmond, E A., & Vande Poll, N. E. (1997). Psychoendocrinological assessment of the menstrual cycle: The relationship between hormones, sexuality, and mood. Archives of Sexual Behavior, 26, 359-382.

Walker, M. L., Gordon, T. E, & Wilson, M. E. (1983). Menstrual-cycle characteristics of seasonally breeding rhesus monkeys. Biology of Reproduction, 29, 841-848.

Wallen, K. (2001). Sex and context: Hormones and primate sexual motivation. Hormones and Behavior, 40, 339-357.

Wallen, K., Winston, L., Gaventa, S., Davis-Dasilva, M., & Collins, D. C. (1984). Periovulatory changes in female sexual behavior and patterns of steroid secretion in group-living rhesus monkeys. Hormones and Behavior, 18, 431-450.

Warner, E, & Bancroft, J. (1988). Mood, sexuality, oral-contraceptives and the menstrual-cycle. Journal of Psychosomatic Research, 32, 417-427.

Werawatgoompa, S., Tankeyoon, M., Kongseripong, R., Chaniyom, K., Virutamasen, E, Dusitsin, N., et al. (1981). Hormonal changes in normal menstrual cycle of Thai women. Contraception, 23, 301-313.

Wilson, M. E., Gordon, T. E, & Collins, D. C. (1982). Variation in ovarian steroids associated with the annual mating period in female rehesus monkeys (Macaca mulatta). Biology of Reproduction, 27, 530-539.

Zehr, J. L., Tannenbaum, P. L., Jones, B., & Wallen, K. (2000). Peak occurrence of female sexual initiation predicts day of conception in rhesus monkeys (Macaca mulatta). Reproductive Fertility and Development, 12, 397-404.

This work was supported by N1H MERIT Award R37 MH41788 to Martha K. McClintock and the Olfactory Research Fund's Tova Fellowship and NIH MD/PhD Training Grant HD-07009 to Suma Jacob. We would like to thank our anonymous reviewers for their helpful comments.

Address correspondence to Martha K. McClintock, Department of Psychology, The University of Chicago, 5730 S. Woodlawn Avenue, Chicago, IL 60637; e-mail: [e-mail].

Manuscript accepted September 24, 2003