Female social and sexual interest across the menstrual cycle: the roles of pain, sleep and hormones

This research is part of a larger study to gain insight into the psychological, social, and behavioral aspects of steroid hormones and the menstrual cycle. A full description of other study measures is provided in Guillermo et al. (in review). Healthy, non-smoking, and regularly cycling women between the ages of 18-30 were recruited from the student population at the University of Nevada, Las Vegas. This study was approved by the University of Nevada, Las Vegas Institutional Review Board and all participants voluntarily provided a written, informed consent prior to enrollment.

Individuals were excluded if they reported diagnosis of mental illness, endocrine disease, substance abuse, medication use or medical conditions that would alter hormone concentrations; this includes the use of OCs among the control group within the last six months of enrollment. Additionally, women were excluded from analysis if they had irregular cycles, as classified by having less than 25 or greater than 40 days and deviating more than 10 days from self-reported average cycle length. Fifty-six women enrolled in the study; however, nine were lost to attrition and seven were excluded due to irregular menstrual cycles. As a result, twenty normally cycling women were assigned to the control group and twenty women using oral contraceptives were assigned to the OC group.

Physical discomfort, sleep disturbance, exercise levels, socio-sexual interests, and hormone values were assessed three times during the course of a single menstrual cycle: early- to mid-follicular phase (T1), mid-cycle (T2), and the mid- to late-luteal phase (T3) corresponding to roughly days 7, 14, and 21 for a 28-day cycle. Specimen collection days and subsequent test times were determined based on individually reported average menstrual cycle length and adjusted for the start date of the next menses after the enrollment interview.

At each session, a questionnaire was administered to measure change in behavior across the menstrual cycle. The questionnaire was derived from sections of several social behavior and attraction questionnaires that included questions regarding physical activity (International Physical Activity Questionnaire; "IPAQ", 2002), sleep patterns (Karolinska Sleep Questionnaire) [17, 18], and interpersonal relations (Interpersonal Attraction Scales) [19, 20]. Questions regarding menstrual cycle related physical discomfort (cramping, bloating, breast tenderness, etc.) were derived from Premenstrual Symptom Daily Rating Scale [21, 22]. A question of general physical activity was included, in which the women were asked about the average amount of vigorous physical activity (minutes/hours per day) over the last three days. Questions rating general attraction, as well as social, physical/sexual, and task-oriented attraction, towards a current or potential significant other were also asked. All questions utilized a 5-point Likert-type scale (i.e. 0 = "not at all" to 4 = extremely). For the purposes of data analyses, questions were categorized by content and analyzed by category. Individual questions and categories are listed in Table 1.

Pre-prandial, morning salivary specimens were collected via expectoration prior to arrival at the testing facility. Hormone specimens were stored in a -25°C freezer until batch analyses by a commercial, CLIA certified laboratory (ZRT Laboratory; Beaverton, OR) using indirect enzyme-linked immunosorbent assay (ELISA) techniques. All hormone testing supplies and analyses were provided and conducted by ZRT Laboratory. Inter- and intra assay coefficients of variation did not exceed 15% for any analyte.

Category scores were based upon the summed scores of individual questions within each category. Attraction questions that reflected a negative interaction with a current or potential significant other were reverse-scaled to maintain consistency in analysis. Within-cycle and between-group differences were calculated using a mixed-model analysis of variance (ANOVA). Assumptions of sphericity for each variable were evaluated using Mauchly's W statistic. If sphericity was violated, adjustments to the significance levels were made using the Greenhouse-Geisser method. All analyses were calculated using the statistical software SPSS 16.0 and PASW 17.0 (formerly SPSS).

Between-group comparisons of demographic features were assessed through Student's t-test. Between-group comparisons of physical discomfort, sleep, exercise and socio-sexual categories were calculated using a series of one-way ANOVAs for each category and by time point. Associations between hormones and outcome variables were calculated using Pearson Product Moment Correlations. Normality was assessed and non-normally distributed values were log10-transformed prior to analysis to meet the normality assumptions of the test. Only testosterone was normally distributed and did not require transformation. Correlations for estriol, which could not be normalized by transformation, were calculated using Spearman's Rho. Post-hoc correlations between individual questions and hormone concentrations were calculated to evaluate more fully the trends in socio-sexual interest.

Table 2 summarizes the demographic data by group. Mean test days were T1 = 5.78 (SD 2.09), T2 = 14.25 (SD 1.77), T3 = 22.50 (SD 2.34) for all groups combined. No differences were found between age, education or body mass index (BMI), whereas a higher proportion of Asian/Pacific Islanders were in the control group in comparison to the OC group. Although the distribution of BMI was not statistically significant, more overweight (25 < BMI < 30) women were naturally cycling (control n = 5; OC n = 2) and more obese (BMI > 30) women were OC users (control n = 1; OC n = 4).

Hormone preparations varied greatly, thus preventing any post-hoc analysis with sufficient statistical power. Of the twenty OC users' pill regimens, 8 were low-estrogen (20-25 μg) versus high-estrogen (30-35 μg) and 5 were multiphasic versus monophasic. Among the different progestin components, 5 women were using OCs containing drospirenone, 5 norgestimate, 4 levonorgestrel, 4 norethindrone, 1 desogestrel, and 1 norgestrel.

As illustrated in Table 3, participants experienced a significantly greater physical discomfort and more disturbed sleep during the follicular phase as compared to any other phase of the menstrual cycle. No cycle-related changes were observed in exercise, or the general, social, and task attraction scales.

Results of between-group analyses are presented in Table 4. No between-group differences in physical discomfort, exercise or sleep were observed. At each phase of the menstrual cycle, women taking an OC reported significantly higher general attraction toward current or potential mates, significantly higher social attraction across all phases and higher physical attraction, particularly during the follicular phase, compared to the control group. No differences were found in task attraction.

As illustrated by Table 5, relationships between individual hormones and the variables measured differed by group and by cycle phase. For the control group, higher estradiol was associated with more physical discomfort at T1 and more perceptions of physical attraction at T3. Higher estriol, however, was associated with less general attraction at T3. None of the other hormones measured were associated with any other variables.

This is in contrast to the OC group where a broader range of hormones was associated with a number of individual indices. For this group, DHEAS was positively associated with general attraction during the follicular phase and exercise levels at mid-cycle and the luteal phase. Follicular estriol concentrations were inversely associated with physical attraction. During the luteal phase, higher estradiol, progesterone, testosterone and DHEAS concentrations were associated with more exercise while cortisol was correlated negatively with physical discomfort.

In an effort to fully understand the role of hormones in socio-sexual behavior, correlations between the individual questions and hormones were calculated (Table 5). Due to the number of correlations, these data should be regarded as trends. Nevertheless, they provide important clues regarding the pattern of hormone-to-behavior relationships in normally cycling and OC-using women. As the table illustrates, associations between hormones and socio-sexual questions, with the exception of progesterone, were negative for both groups during the follicular phase. At mid-cycle, higher cortisol was associated with "enjoying the company of the other" for the control group, whereas higher estradiol and DHEAS concentrations were linked positively to task attraction in the OC group. During the luteal phase, higher estradiol and estriol were associated with sexual interest questions in the control group, whereas higher progesterone was associated with more social attraction and better work interactions in the OC group. In addition, higher luteal DHEAS meant lower social and physical attraction for the OC group.