Winnifred B. Cutler* Celso Ramon Garcia** and Abba M. Krieger ***
*Department of Obstetrics and Gynecology, Hospital of the University of Pennsylvania
**Division of Human Reproduction, Department of Obstetrics and Gynecology, Hospital of the University of Pennsylvania
***Department of Statistics, The Wharton School of the University of Pennsylvania, Philadelphia, PA 19104
Hormones and Behavior 14, 163-172 (1980).
© Copyright 1980 by Academic Press, Inc.
*Please click on bold print within text below for details and/or studies in full.
*Please click here for more on pheromones and books from Athena Institute
In an initial study of 248 university students prospective double-blind monitoring of sexual behavior and menses onsets showed that sporadic (less than regular weekly) sexual activity was associated with aberrant menstrual cycle lengths. This study examines the sample of sporadically active women for behavioral correlates of aberrant cycle lengths. It is shown that women reporting a high frequency of sexual activity (though in a sporadic pattern) showed a higher incidence of aberrant cycle lengths than women reporting a low frequency of sporadic coitus. Stable ongoing sexual behavior patterns appear to associate with a different endocrine background than less stable, sporadic sexual patterns.
Recent human studies of sexual behavior show some of the associations with reproductive state that have been reported in other mammals. That endocrine states correlate with sex behavior changes has been studied in humans by following the sexual behavior or women across the menstrual cycle. Recent results have been essentially consistent. Udry and Morris (1968) reported an increase in coital activity at midcycle, although their data 9 years later were somewhat less supportive (Morris, Udry, and Underwood, 1977). Both coital and orgasms frequencies were reported to increase during the late follicular (early preovulatory) phase of the cycle (Erickson, 1977). Furthermore, a midcycle increase in female-initiated sexual activity was recently reported in both human (Adams, Gold, and Bur, 1978) and rhesus (Michael, 1979).
The possibility that changes in sexual behavior produce changes in endocrine state has also been investigated. Jochle (1975) suggested the existence of a low incidence of reflex ovulation in response to coitus for all the follicular phase days of the cycle by citing a series of anecdotal reports of single coitus producing pregnancy in women. McClintock (1971) showed that college women who saw men frequently, maintained shorter menstrual cycles than their classmates (in an all female college) who saw men less frequently. Although two research groups were unable to demonstrate an LH response to coitus (Stearns, Winter, and Faiman, 1973; Davidson and Trupin, 1975), the time course of a possible postcoital LH response has not been resolved and could have affected these results, especially so since an LH response to erotic films was demonstrated in college men (La Ferla, Anderson, and Schalch., 1976).
It was recently reported that the temporal pattern of sexual behavior correlates with the length of the menstrual cycle in both infertile and fertile women (Cutler, Garcia, and Krieger, 1979b). Regular weekly heterosexual activity was associated with menstrual cycles of classic ovulatory cycle lengths (26 to 33 days). Women showing less that regular weekly heterosexual activity, as well as heterosexually inactive women, more often exhibited menstrual cycles of either short or long lengths which others (Treloar, Boynton, Behn and Brown, 1967; Strott, Cargille, Ross, and Lipsett, 1970; Dodson, MacNaughton, and Coutts, 1976; Saxena, Poshyachinda, and Dustin, 1976; Garcia and Rosenfeld, 1977; Vollman, 1977) , have found to be characteristic of anovulatory or otherwise subfertile cycles. See Fig. 1. The data from later studies showed that a particular type of infertility - the short hyperthermic (luteal) phase - was associated with sporadic sexual activity among infertility patients. The short luteal phase is characteristic of “luteal insufficiency" a recognized pathologic syndrome (Cutler et al., 1979c).
Figure 1. Sexual behavior frequency and menstrual cycle length 248 university students
Half of the sporadically active college women showed aberrant cycles; half showed classic cycles, and it is now asked whether a pattern of sexual behavior can distinguish the sporadic women with aberrant cycles from the sporadic women with classic cycle lengths. Since sporadic (less than weekly) sexual activity was associated with aberrant cycle lengths, it seemed possible that a sporadic temporal pattern itself, might disrupt the cycle; and if so, a woman might have an increased likelihood of a subfertile cycle if she had an increased frequency of disruptive behavior. This report asks: Do the sporadically active women who report high frequencies of coital activity have a higher incidence of aberrant cycles than the sporadically active women with lower frequencies of coital behavior?
Specific details of the prospective double-blind data collection of sexual and menstrual occurrences are reported elsewhere (Cutler, Garcia, and Kreiger, 1979b) but a brief overview is presented. The sample is comprised of 248 gynecologically mature (menstruating >7 years) college women who used neither oral contraceptives nor intrauterine devices. The women ranged in age from 19 to 22, were unmarried, nonroommates, nulliparous, and white. Each onset of menstruation and coital day during a 14-week fall semester was prospectively monitored by data collectors who were female contemporaries of the subjects. Subjects did not report on the number of sexual events per day. Coital behavior was categorized as none, weekly (the subject had sex at least once every week throughout 14 weeks), or sporadic (less than regular weekly sexual activity). In addition, the number of days of coitus during the 14 week span was totaled. The assignment to the weekly category, therefore, reflects a temporally stable pattern of behavior in which coitus occurs at least 1 day in every single week throughout the 14-week semester. Thus the term “sporadic” covered a wide range of sexual frequencies - from a minimum of 1 day of coitus during 14 weeks to a theoretical minimum of 13 highly active weeks (provided 1 full week of sexual abstention occurred). Others have also shown that similar uses of weekly categorization are fruitful: in men (higher) weekly ejaculation frequencies associate with (higher) sperm quantity and quality (MacLeod and Gold, 1953) while among women, higher weekly frequencies of “association with males” (sexual activity undefined) were associated with shorter menstrual cycles than less frequent association with males (McClintock, 1971).
Figure 2 shows the total number of days of coital exposure of the 23 weekly active women. Each dot shows the mean menstrual cycle length (x axis) of one woman and her total number of coital days (y axis) during the semester. It is noted that the total number of coital days for these weekly subjects ranged between 17 and 68 and that no woman had an average cycle length shorter than 27 days. Furthermore, the raw data of individual cycles showed that there were no aberrant short cycles among these coitally active women and relatively fewer that lasted more than 33 days when compared with the sporadic women. As women could have between one and four cycles during the study period the first, last, or mean cycle length has been used in the statistical analyses. 1 Similar results were obtained with each measure (Cutler et al., 1979b)
Figure 3 shows the total number of coital days of the 89 sporadic women. Each dot shows the mean menstrual cycle length (x axis) of one woman and her total number of coital days (y axis). It is noted that the total number of coital days for these sporadic subjects ranged between 1 and 51 and a sizeable portion (half) of these 89 women show aberrant short and long cycles.
1. It may, at first appear that analysis of variance statistics are the appropriate analytical measures for these data. The following explanation should clarify the issue. In general, even if the statistical procedures did permit an evaluation of variance in cycle length, this would not be the issue we are addressing in this report. For the purpose of this report, we are demonstrating a relationship between the experience of aberrant lengths and a particular pattern of behavior. A variance measure less accurately reflects incidence of subfertility since it does not directly address the issue of cycle length. As examples consider the following cases:
(1) A woman ha a 45- and 47-day cycle. Both are likely to be subfertile, yet carry a low variance score of 2.
(2) A woman has 16-, 17-, and 18-day cycles. All are subfertile (see Vollman, 1977) yet carry a low variance score of 1.
(3) A woman has 26, -28, and 33-day cycles. Her mean of 29 correctly reflects a fertile range but her variance of 13 is high.
The only danger of our use of mean cycle would be cases where a woman had both extremes (very short and very long) in such a combination to yield a mean in the classic range from 26 to 33 days. In such a case, an aberrant cycle would fail to be properly identified. Such incidence is rare. Among the weekly women, there was NO case of this type. Among the 89 sporadic women by classic-type means there were three as follows:
Subject No Cycle Lengths Cycle Mean Coital Total 0603 40,19,33 30.67 16 0813 42,26,14 27.33 9 0811 30,21,23 28.00 14
All three have aberrant cycles with below 17 days of coital activity. One could be ultraconservative, recast the table and redo the X2. The new table would have cells as follows:
3 14 Instead 3 14 14 24 41 31
However, the statistic does not change to the second decimal place and the significance is unaltered.
Since 17 was the lowest coital frequency of any weekly subject, this level was set as a threshold to evaluate whether the sporadically active women who attained this threshold tended to show an increased incidence of aberrant cycle lengths. Table 1 arrays the distribution of the 89 sporadically active women with respect to cycle length and coital threshold. It is noted that 31 (43%) of the 72 below-threshold sporadics show aberrant cycles. It is further noted that of the 17 women with high coital frequency, 14 (82%) show aberrant cycle lengths. Thus, of these 17 most active sporadic women, only 3 (18%) had cycles of classic ovulatory length. Extreme cycles are associated with frequent (above threshold) sexual behavior in sporadic women (X2 + 8.48, p <0.002). The Wilcoxin rank test which assigns no particular threshold value to the “high” coital frequency was also applied to compare the 44 classic length cyclers to the 45 aberrant cyclers to see if these two sporadic subgroups showed statistically different mean coital totals. With no preselected threshold the trend remains (z= 1.44, p<0.075).
A scan of the raw data - to reveal possible clustering patterns of coital spans within the semester - revealed that there was no major clustering of coital activity on narrow segments of the study period. No single week (s) tended to be uniquely without sexual activity.
Sporadically active women with classic-type cycles (26-33) might have shown a stable pattern of sexual activity, albeit with a minimum repeating span of greater than weekly duration. Accordingly the pattern of sexual activity of the sporadically active women who showed classic-type cycles was inspected for possible consistency of ANY pattern throughout the 14-week study. All data were rearrayed to expose duration between and days of coitus as well as to show menstruation occurrence. Inspection of these arrays, with particular attention focused at scanning for patterning among the sporadic women with classic-type cycles revealed no consistency in the sexual behavior pattern of these women. Furthermore, when average intervals between days of coital activity (whether measured individually or as operationally defined “coital bouts”) was computed, this way of looking revealed no differences between classic or aberrant cyclers among sporadic women with similar overall coital frequencies. Thus, inspection of these data revealed no distinguishing patterns of behavior which might separate the classic from the aberrant cycle except for the overall total number of days of coital activity during the 14-week study. Likewise, a comparison of the data arrays of weekly and sporadically active women with > 16 days of coitus in the semester also was without any obvious difference in average intervals between coitions. Inspection of Fig. 2 might suggest that high coital frequency among weekly active women increases the likelihood of aberrantly long cycles. While there are too few data to permit meaningful analysis, inspection of the arrays was performed. The inspection revealed no difference in patterns for either the aberrant cyclers or the women with high coital totals when each subgroup was compared with the remaining weekly active women.
Thus the total quantity of coital days which associated with classic cycle lengths in the weekly group was not sufficient as an associate of classic (26-33) cycle lengths in the sporadic group. In fact, high coital frequency among the sporadically active women associated with aberrant menstrual cycle lengths.
It is noted that the occurrence of 17 days or more of coital activity among the weekly active women associates with cycle lengths which are mostly within the classic range of 26 to 33 days; while that of 17 days or more of coital activity among the sporadic women associates with cycles which are mostly outside the classic range 26 to 33 days. While these data do not permit a determination of causality, it is tempting to consider a circular relationship where alteration of the behavior from irregular to regular sexual activity might produce classic type cycles AND alteration in physiology from aberrant to classic-type cycles [ as has been reflected in experimental manipulation on rhesus monkeys (Michael, Richter, Cain, and Bonsall, 1978)], might produce sexual behavior changes. Thus, one could conceive of a closed loop which could be entered at either site to yield a prediction of outcome.
The finding that sporadic sexual behavior is associated with the type of cycles that others have related to anovulatory cycles suggests that coital activity throughout the entire cycle serves some function. In fact, these data showed that luteal phase activity was just as necessary as follicular phase activity to yield the association shown (Cutler et al., 1979b, c). Thus, there may be a physiological importance for nonovulatory coital activity. Primates are unique among mammals for their copulation at nonovulatory times. Perhaps primate sexual activity, in contrast with that of other mammals, serves to prime the reproductive system as well as impregnate it. Support for this hypothesis derives from a recent report that delayed age at first coitus associated with an increased incidence of subsequent infertility a number of years later (Cutler et al., 1979a). In that report the pubertal ages were the same (12.44 or 12.45) for fertile and infertile women.
The sporadically active women of this study frequently mentioned that their sporadic activity tended to result from disruptions in relationships; either breaking up with one's sexual partner or vacations from campus which temporarily separated the sexual partners., Subjects were not asked to report on their basal body temperatures or on the numbers of their sexual partners; nor were they requested to donate blood for sera analysis. Since this was the first human sexual behavior study ever undertaken among undergraduates in recent times at the University of Pennsylvania, it seemed prudent to ensure maximum participation by eliminating any possibly onerous requirements to either the subjects or their data collectors. Such information would be highly desirable for future studies.
Regardless of the direction of causality, it does appear that a weekly stable ongoing sexual relationship is associated with a very different endocrine background, as menstrual cycle length reflects (Vollman, 1977), than a sporadic and presumably less stable one. In fact, for women whose sexual activity occurred on a sporadic (less than every week) basis, high frequency in coital days was associated with a high frequency of aberrant cycle lengths.
One value of these results lie in the possibility that they may lead to a demonstration of a neuroendocrine feedback loop in which sexual activity directly alters endocrine levels or patterns, as well as the better known mammalian phenomenon of endocrine function directly altering behaviors and/or thresholds.
The mechanisms by which sexual behavior might exert control over the menstrual cycle cannot be resolved with the data of this report. That some combination of genital, emotional, and pheromonal afferent stimulation could be involved derives support from a large body of literature showing that various combinations of all three produce neuroendocrine responses to affect reproductive processes in a variety of non-human animals. On women topical application of synthetic aliphatic acids, prepared to mimic the naturally occurring vaginal ones, did under certain conditions, increase coital frequency (Morris and Udry, 1978). Once could speculate that a female response to male pheromones (perhaps produced in seminal fluid or elsewhere) might likewise, influence female sexual behavior and/or endocrine responses.
Likewise, one could speculate that a difference in attractiveness resulting from some form of hormone -dependent patterns (i.e. , secondary sexual characteristic or hormone-dependent sexual motivation) might be reflected in increased likelihood for regular weekly sexual behavior.
Since there has been some suggestion that LH is needed to maintain the corpus luteum (Casper and Yen, 1979; Vande Weile, Bonumil, Dyrenfruth, Ferin, Jewelwicz, Warren, Rizkallah, and Mikhail, 1970) and that sexual activity may produce LH (see introduction) one might suggest the following speculative scheme: Perhaps luteal phase sexual activity serves to stimulate sufficient LH production to maintain the corpus luteum. Likewise follicular phase activity serves to prime the steroid-gonadotropin secreting system sufficiently to ensure the LH surge which preceeds ovulation. Since the stimulus for LH secretion in the human and rhesus is estrogen (Yen and Lein, 1976; Knobil, Plant, and Wildt, 1980) it may be that sexual activity in humans stimulates estrogen secretion. Alternatively a woman with levels of estrogen sufficient to maintain this complex cycle might fulfill the attractivity conditions suggested above. The issue is clearly moot.
The inability of the data arrays of this 14-week study to reveal repeating cycles of behavior among the classic length sporadic women could reflect too short a time period for adequate analysis. It would be useful for future investigations to dusty behavior over longer intervals. For this, a noncollege sample would seem necessary.
Alternatively, one might view these results from a broader prospective and ask the following questions. Why does the “week” seem to provide the best of interval for the marking of human reproduction patterns and fertility? Does the civilization which marks off a sabbath each 7 days help synchronize a body rhythm at this pace? Would a society which counts time on a 5- or 6-day week show an appropriate change in the minimum repeat necessary for maintaining the cycle? Or was the establishment and maintenance of the 'holy cycle' of 7 days, a natural selection process? Did those cultures which marked time this way survive to reproduce? These questions intrigue us. Perhaps investigators may eventually focus on them.
Studies which will attempt to answer some of the resulting questions of mechanism are being planned in order to explain why increased incidence of sporadic coital behavior associate with increased incidences of subfertile cycles.
We thank Theresa Allen and Eliot Stellar for their contributions to this manuscript, as well as Jerre Levy for initial encouragement of menstrual cycle studies.