The BioMeter: An in vivo tissue biosensor for fertility awareness and possible additional applications. Part 1.



A review of the fertility window problem and of the commercially available technologies addressing the problem



Vaclav Kirsner, Ph.D., BioSense Corporation



Introduction


In this first part of the report, we critically review the problem of the determination of the fertility window, and the commercially available technologies addressing the problem. The BioMeter and how it operates is then reported on in part 2.

The development of the BioMeter was undertaken because of the shortcomings of the existing approaches to fertility status determination. The fundamental scientific advantage of the BioMeter is that it detects the fertility status directly - by means of the biological valve actively involved in the regulation of reproduction, the cervix uteri - rather than through some indirect indicator such as those involved in the other techniques of fertility monitoring. The nature of the fertility indicators is of critical importance because the fertility window is only a few days wide due to the limited lifetime of the sperm, and even more critical due to the very short lifetime of the ovum. It is also well understood, although often ignored in practice, that the position of the ovulation day within the menstrual cycle is usually quite variable, with the cycle length changing from cycle to cycle in the same woman, as well as within a population.

The fertilizable lifetimes of the human egg and sperm are not known with certainty. The ovum is thought to be fertilizable for only about 8 to 12 hours after ovulation, though some estimates are for up to 24 hours after ovulation. According to Speroff, Glass and Kase's monograph on gynecologic endocrinology [1], "equally uncertain is knowledge of the fertilizable lifespan of human sperm. The most common estimate is 48 - 72 hours", which means that the sperm is thought to be capable of fertilizing the egg for up to at most 3 days after being deposited in the vagina. This is consistent with the "Gamete Transport" chapter by Croxatto in the authoritative compendium edited by Adashi, Rock and Rosenwaks [2]. Ovulation must therefore be anticipated more than 3 days in advance, and it must be detected with sufficient accuracy.

This has not been possible to achieve with the available personal diagnostic methods, and it has been a major problem even in studies that relied on laboratory procedures of fertility determination. As a consequence, the fertility window - which translates as the period of abstinence in natural family planning (NFP) - has been postulated to be much wider than necessary, in order to minimize the chances of unwanted fertilization. In the practice of NFP, as employed predominantly by Catholic couples, the required period of abstinence can be longer than 10 days. This excessive length, along with the perceived lack of reliability, has contributed to the lack of popularity of NFP in countries with easy access to other means of birth control such as chemical contraception and sterilization.

This is so in the U.S., even though the Pill is the most popular contraceptive only among the youngest women [1]. Closer to 30 and after 30 years of age, women are more concerned about chemical interference with the functioning of their bodies, and pay more attention to other contraceptive choices, including fertility awareness. This is also the age group where infertility or sub-fertility becomes a difficult problem (believed to be caused mainly by the postponement of pregnancy till later years) which the BioMeter can help to deal with. The timing of procedures in the management of impaired fertility is, of course, the other application where fertility status determination is essential.

Other electronic devices aspire to being useful for personal fertility monitoring, and these will be critically reviewed below in relation to the BioMeter.





The fertility window problem: The variable boundaries of the intermittent fertile state in the human female


In this section we focus on the timing problem as it pertains to fertility awareness and/or to natural family planning as well as to infertility treatments. Fertility awareness could be a generic term but the phrase tends to be used to describe the birth control practice whereby some method of contraception is applied during the fertile days of the menstrual cycle. That is in contrast with natural family planning whereby sexual abstinence during the fertile days is used as a means of birth control. Yet another descriptive term is the rhythm method (or the calendar method), according to which the intermittent fertility window used to be calculated based on the assumption that the cycle length remains constant, often further assumed to be of the idealized 28 days duration.

This misconception tends to be carried over to areas such as the management of premenstrual syndrome or the estimation of the expected date of birth: the healthcare provider assumes the 28 day cycle length and, in the case of PMS, will time the treatment as though ovulation took place 14 days after the start of menstrual bleeding. Yet, it is rather rare that a woman will ovulate on day 14 and have a constant length of cycle, let alone one of 28 days. Literature indicates 12% or 15% incidence of the 28 days long cycle in a population [1].


For example, in the Marquette University pilot study discussed in Part 2, the cycle lengths ranged from 23 to 35 days in a group of 21 cycles recorded by 10 women. The estimated timing of ovulation ranged from day 9 to day 23 (or from 10 to 24, depending on the method of detection), and only 2 of the 10 women appeared to ovulate on the same day of cycle in two successive menstrual cycles (by peak mucus and LH kit data but not by the data yielded by a BioMeter prototype). This outcome is quite consistent with the results of large studies focused on cycle length investigation [3] [4] [5]. The one subject who monitored four successive menstrual cycles, a 41 year-old nulliparous woman, produced the following data for the estimate of the ovulation day in the successive cycles:
14, 13, 15, 14 by LH kit,
15, 14, 16, 13 by peak mucus, and
18, 13, 18, 15 by BioMeter prototype.

Let us note the following about these data. The spread of ovulation timing (2 to 5 days, depending on method) is comparable with or greater than the fertilizable lifetime of the sperm, as is the uncertainty due to the different methods. The LH kit and peak mucus data are within one day of each other in all four cycles but the BioMeter data is within one day of one or the other only in two of those cycles (it either coincides or follows 1 day later). In the other two instances, the BioMeter indicates a significantly delayed ovulation. At this point, we have not introduced any data to make sense of these discrepancies but, in the following sections, the reader should get an understanding that the BioMeter is capable of detecting such delays between the hormonal stimuli and the actual occurrence of ovulation. This is because the BioMeter is a monitor of end-organ effects.

The LH kit data is a urine-derived representation of the brain signaling to the ovary that ovulation should proceed. The peak mucus represents the ovarian estrogen-driven change in the quality of mucus at around the time of ovulation but it is a subjective observable and one that is not sufficiently tightly coupled with the ovulation event. Peak mucus can be observed from 3 days before to 3 days after ovulation, as has been documented by experts both within and outside of natural family planning [3, 4, 5]. We can do little with data such as those above in the absence of ultrasound visualization of the rupture of the follicle, which is the best technique of direct confirmation of ovulation (about 80% reliable due to possible follicle rupture without egg release). Neither the LH kit nor the peak mucus is any gold standard for ovulation detection.

A 1986 symposium on ovulation prediction in the treatment of infertility covered all the phenomena known to be associated with ovulation [9]. Moghissi, who discussed more than 20 measurable parameters that vary during the menstrual cycle, stated the following [8]: "Mid-cycle mucorrhea, ferning, spinnbarkeit and lowered cell content and viscosity of cervical mucus are used commonly in ovulation detection and as an index of the estrogenic response of cervical epithelium. However, these changes extend over several days ... (These changes) do not necessarily indicate ovulation, and are merely an index of the optimal amount of circulating estrogen...". It must also be pointed out that the literature on ovulation prediction suffers from the fact that the event of ovulation is usually estimated rather than determined or confirmed to occur by definitive means.

Against this background we recall that a 1985 study [10] by the World Health Organization showed that utilization of the steroid hormone data to estimate the day of ovulation resulted in a far from satisfactory outcome. Using the steroid hormone ratio resulted in fertility window of 10 +/- 2 days, and agreed with the LH test in 74% of cases. Ten years later, another frequently cited publication [11] assumed that the steroid hormone ratio is a 100%-reliable correlate of the LH surge, and concluded that the fertility window is 6 days wide with a sharp cutoff at ovulation. Assuming the exact correspondence, and therefore tight coupling between the hormonal signals and ovulation, resulted in the highly publicized novel finding that no pregnancies resulted from intercourse after ovulation (not even on day +1).

All efforts to develop a practical fertility awareness (or NFP) method have been based on the endocrinologically defined and WHO-sanctioned window of fertility, defined in terms of the cyclic profiles of estrogen, progesterone, luteinizing hormone (LH) and follicle stimulating hormone (FSH). Ninety percent of conceptions have long ago been estimated to occur within a 6-day window, from day -3 with respect to the peak of urinary LH to day +2, but this estimate had an important caveat. The caveat was the recognition that "there is a minimum error of +/- 12 hours in this estimate, which in practice may be +/- 24 hours" [10], because the LH is an imperfect reference.

The 1995 New England Journal of Medicine publication [11] reported on work carried out between 1982 and 1985, i.e., at about the time of the WHO study [10]. The Wilcox et al. team performed statistical analysis of data on hormone measurements of frozen samples of urine and data from women's journals of sexual activity "from the time they stopped using birth control until the eighth week of clinical pregnancy or up to six months if no pregnancy was clinically evident". There was no other effort to determine ovulation besides the quantification of estrogen and progesterone metabolites in urine. The authors assumed that the estrogen/progesterone ratio (E2/P) corresponds to the LH surge, which was in turn believed to "correspond approximately with the day of ovulation".

The Wilcox et al. paper [11] contradicts certain findings of a 1992 publication by John T. France et al. [12] of a study with the same goal but substantially different protocol. The main distinguishing features of this 1992 publication are that it is based strictly on data whereby only one coitus per fertile period occurred, and that three different markers were used to estimate the time of ovulation.

The stringency of the study design by France et al. went so far as to exclude 29% of pregnancies from the birth sex ratio evaluation in terms of timing of conception with respect to ovulation, because of more than one act of intercourse during the fertile period. Interestingly, the birth sex ratio was 0.50 for this excluded group but far from 0.50 for the good study population. This difference should be taken into account in the design of future studies. The results of the France et al. study were then as follows: Of the 34 male infants born, 65% were conceived 2 to 5 days before ovulation, and 71% of the born girls were conceived from intercourse timed between 1 day before to 1 day after the estimated time of ovulation. However, there was a great uncertainty about the actual ovulation day because in only 9% of the cases did the three ovulation markers agree with each other. In 68% of the cycles, agreement was within +/-1 day. The peak cervical mucus marker was one while the other two markers were the onset - not the peak - of the LH surge, and the basal body temperature rise.

In the paper by France et al., the fetal sex distribution plot on the peak mucus-referenced time scale reveals two remarkably well defined bell-shaped curves. The female births peak on day –1, while the male births peak on day -3. The male births distribution curve is also notably wider than the female distribution curve, on both sides of the apex. The fetal sex distribution plots based on the other two markers do not present such a pattern as convincingly, although one can be discerned in the BBT plot. The BBT birth distributions would seem shifted in the positive direction (to the right) with respect to the peak mucus plot: the male birth peak is shifted by 2 days (from day -3 to day -1) and the female birth peak by one day (from day -1 to day 0). While the amplitudes of both peaks are lower than in the mucus graph (9 males vs. 10, and 7 females vs. 9), clearly because some of the births are now included in neighboring counts, the male distribution curve can again be suspected to be wider than the female curve, although this is much less clear than on the peak mucus time scale. It is notable, however, that the female birth distribution declines rapidly on the positive (right) side even in the least well pronounced distribution pattern yielded by the data referenced to the start of the LH peak.

Comparing the three birth distribution plots brings out clearly the importance of the quality of the ovulation marker for the outcome of such studies. This, rather than any conclusion about the feasibility or otherwise of fetal sex pre-selection, can be read into the outcome of the France et al. study.

We note one possibly meaningful qualitative agreement between the data in the 1992 paper by France et al. and the 1995 paper by Wilcox et al., keeping in mind that the 1992 data are from 55 pregnancies (and births) while the 1995 data are based on 34 pregnancies. These 34 pregnancies resulted from 129 menstrual cycles in which intercourse occurred on a single day during the fertile period, as in the other study. Remarkably, if the sums of male and female births from the France et al. data are plotted against cycle day with respect to ovulation (as estimated by peak mucus), a pattern similar to that found by Wilcox et al. is revealed. Both show three days of comparatively high conception rates; Wilcox from day -2 to day 0 while France from day -3 to day -1.

In the data of both studies, an overwhelming majority of pregnancy or birth counts is concentrated within these three days (82% in the Wilcox et al. data and 65% in the France et al. data), and only a small percentage of the counts occurs within the skewed boundaries of the respective distributions. (Skewed from the respective high pregnancy/birth count or probability of conception to zero count or probability.) Also, the boundaries of both patterns are skewed in the same manner: Both distribution curves fall off slowly in the pre-ovulatory phase (both decline over 4 days), and they fall off rapidly after ovulation, although only the 1995 Wilcox et al. data fall off to zero immediately after the estimated day of ovulation. In view of the one-day uncertainty in the underlying estimate of ovulation (from E2/P data), this sharp cutoff may or may not be real. Furthermore, despite the fact that the E2/P ratio drops rapidly upon ovulation (as a result of luteinization of the dominant follicle), it is not clear why "intercourse that was recorded on a given morning was assumed to have occurred the previous day". That assumption could clearly put any day +1 scores into the count of day 0 scores, and thus lead to the unprecedented outcome of the surprisingly sharp cutoff at ovulation.

A similar criticism applies to the data of France et al., based on the uncertainty of +/- 3 days in the peak mucus method. Their birth distribution statistics after ovulation are of dubious validity, comprising as they do of 6 births on day +1, plus a single birth on day +2. The discrepancies between the three timing methods (onset of LH peak, peak mucus, and BBT) are particularly worrisome, because there are significantly fewer post-ovulation birth counts on the other two time scales. On the LH time scale there are merely single birth counts on each of the 3 post-ovulation days, or a 57% lower post-ovulation score. There are also fewer post-ovulation births on the BBT time scale, namely, 5 post-ovulation birth counts compared to the 7 births on the peak mucus time scale, or a 29% lower score.

In fact, the France et al. data are such that day +2 is a "fertile" day based on a single birth on the LH-onset time scale, no birth on the peak mucus time scale, and two births on the BBT time scale - all out of 55 births total.

The most significant quantitative difference between the overall birth distributions in the two studies is in their widths. The 1992 birth distribution by France et al. is 9, 9, and 10 days wide, depending on the method of estimating ovulation, and with the largest value of 10 resulting on the start-of-the-LH-peak time scale. In 1992, this width of the fertility window was in agreement with the then recently WHO-generated figure of 10 +/- 2 days. The 1995 birth distribution by Wilcox et al. is only 6 days wide, and this was enthusiastically celebrated by the popular media.

It is the contention of this writer that it is quite possible that in fact the real fertility window is even less than 6 days wide. This is because the low birth counts on the extremes of the birth distribution curves in the discussed publications are probably experimental artifacts resulting from inaccurate determination of ovulation. In the discussed studies, ovulation was merely estimated, not detected as ultrasound-visualized follicle rupture, with further confirmation wherever possible (e.g., by cul-de-sac fluid culdocentesis data). As a result, we see such data as three births on day -6 in the France et al. publication, which is difficult (if not impossible) to reconcile with the most commonly accepted 2 to 3 days of fertilizable lifetime of the sperm [1, 2]. If these three births were data point outliers that in fact belong to one or another day count within 3 days of day -6 (the uncertainty in the peak mucus method), this and other such adjustments could considerably alter the birth distribution pattern.

A similar consideration applies to the Wilcox et al. publication of 1995. Their data are based on a method of estimating ovulation that has a declared uncertainty of one day, as per the WHO study [10] discussed above. Their birth distribution data, which they selected from the single-intercourse data set (Fig. 2 in [11]), contains two groups of birth counts (hence proposed fertile days): the group of high conception probability referred to above as occurring on days 0, -1, and -2, and the other group with considerably lower conception probabilities (on days -3, -4, and -5). The data in their Table 1 reveal that the second group is made of only 6 pregnancies versus 28 pregnancies in the first group.

In fact, the Wilcox et al. conclusion that day -5 is a "fertile day" is based on recording a single pregnancy on day -5, out of 12 attempts, i.e., out of 12 different women's cycles where intercourse was recorded only on this day. This data point could well be an outlier due to experimental error, in view of the known uncertainty in the E2/P method of estimating ovulation (consistent with the difficulty to reconcile 5 days before ovulation with at most 3 days of fertilizable lifetime of the sperm). Thus, the validity of declaring day -5 a "fertile day" is in doubt, in this writer's opinion.

A possible interpretation of the qualitative agreement in the distribution profiles of the two studies is as follows. There is an undeniable 3-day fertility window, during which period there is a high probability of intercourse resulting in conception. The 3-day width of the fertility window is consistent with the most common lifetime estimate of the human gametes [1]. The position of this fertility window with respect to ovulation depends on the method of determining ovulation. With an unequivocal detection of the ovulation event, the fertility window should be timed equally unequivocally.

The appearance of the 3-day "skewed boundary" in the pre-ovulation data of both the discussed studies is probably the result of the fact that the respective methods of estimating ovulation are based on hormonal signals that precede ovulation. This is true of the LH as well as of the steroid hormones (and therefore also of the steroid-controlled peak mucus). Any delay between the hormonal input and the actual occurrence of ovulation cannot be detected in these data, and artifacts can and will arise. The apparent pre-ovulation births in the BBT data cannot be explained in this manner because the BBT rise is not supposed to occur before ovulation, but the BBT is well known to be unreliable.

Data of both studies show a faster decline of the post-ovulation "skewed boundary" but there is a significant quantitative difference in the duration of the decline (2 days vs. 0 days). This difference is likely the consequence of the respective methods of estimating ovulation in that the E2/P is apparently more tightly correlated with the LH surge and with ovulation than are the other three parameters used by France et al. (including the onset of the urinary LH rise). The fact that the Wilcox et al. data suggest an immediate cut-off as opposed to a finite decline might be related to the one-day shift of the respective high fertility windows (Wilcox: days -2 to 0 vs. France: days -3 to -1). A wider skew would logically arise from the methods with larger margins of error as compared to the +/- 1 day uncertainty in the E2/P method. This would prevent an identical outcome even in the absence of the Wilcox et al. assumption that a given morning's score counts as previous day's score.

Future studies ought to aim to find out if the apparent qualitative agreement between the two studies can be confirmed, and if so what the meaning of it is. The apparent qualitative agreement is the fast fall-off of the post-ovulation conception scores as compared to the pre-ovulation fall-off. Should the relatively fast post-ovulation fall-off be confirmed, then the possible difference between the male and female birth statistics, as suggested by the France et al. study, should be investigated in this context.

It is the writer's contention that the France et al. protocol should be applied - after modifications such as obtaining birth or pregnancy scores from identical number of attempts for each tested day - in a study using the BioMeter to anticipate and detect ovulation. The study should establish the fertile window against that data, correlated with ultrasound. It is imperative to calibrate the BioMeter against the ultrasound gold standard, as should be done with any technique introduced as a tool for fertility awareness or natural family planning.

A possible formulation of the goal of the project is to establish whether the real or practical fertile window is in fact only 3 days wide (and which, if any, of the two suggested positions of the fertile window with respect to ovulation is the true one), or whether the fertile window is 4 days wide (or more). This could arise as a result of some conceptual overlap of the two suggested 3-day periods (days -2 to 0 and -3 to -1). The latter might result possibly as an experimental error superimposed on the theoretical window of fertility defined by the lifetimes of the gametes.






A critical review of commercially available fertility diagnosis technologies


There are two kinds of widely used fertility diagnosis products with a well established presence in the marketplace. One is the thermometer, either generic or customized for the so-called basal body temperature measurement, and the other is the chemical kit for the detection of the LH (luteinizing hormone) in the woman's urine. None of these can predict ovulation sufficiently well ahead of its occurrence so as to be useful for the birth control methods of natural family planning or fertility awareness. There are also two other electronic technologies that are less widely used, and these devices will also be critically reviewed below (the Cue from Zetek Inc. and the more recently introduced ClearPlan Easy from Unipath Ltd.). This review is not intended to be comprehensive but rather to outline the practical consequences associated with the problems discussed in the preceding section, and to point out where the BioMeter improves on the older techniques. A number of the personal fertility diagnosis products may be found on the web site http://www.fertilityproducts.com.

The thermometer approach to fertility diagnosis was developed several decades ago as the only self-monitoring technique available at the time. Although attempts have been made to go beyond its accepted capability (notably by Fertil-A-Chron and their BioSelf system), temperature monitoring can only indicate that ovulation has already occurred. The basal body temperature cannot predict ovulation. Even the retrospective timing of ovulation from temperature records is uncertain since the post-ovulation temperature rise may be delayed by several days. This is because the small increase in the woman's temperature after ovulation is an indirect indicator, driven by a post-ovulation increase of a hormone concentration in blood circulation. It is suspected that, while the woman's temperature-regulating mechanism is sensitive to the sex hormone progesterone, the primary modulator of temperature regulation is norepinephrine, and that progesterone triggers its release. Many aspects of life-style affect the body temperature, and each temperature reading may be influenced by factors other than the phase of the menstrual cycle.

Not surprisingly, the BBT was the reference method we used at the inception of the BioMeter project, which goes back so far as to have taken place before the introduction of the LH kits. The development of the personal fertility tool based on the urinary LH was vigorously under way at that time, and it appeared to be an obstacle for gaining support for the different approach to the problem. It should be said that in our early work the BBT rise was always observed to occur after the newly discovered ovulation marker of the technique under development, and it would have been confusing had this not been the case. It is equally not surprising that the France et al. team, whose work was discussed in the previous section, utilized the BBT method even as recently as they did in the early nineties, because of its easy availability and low cost.

Rather than result in a reliable fertility monitor, the many published basal body temperature studies have demonstrated that menstrual cycle regularity is a myth. Most women experience changes of more than five days in the length of the menstrual cycle. Less than 1% of women would be found with no variation at all, even for short sequences of only a few menstrual cycles [3]. Two longitudinal studies are probably the most frequently cited sources of data on this subject, published by Vollman [4] and by Troelar et al. [5] (discussed in both [1], page 219 and [3]).

The cycle-length variability affects even the use of the more recently introduced type of self-diagnostic product, the LH kit for urine analysis. This is because the LH kit provides a set of chemical reagents for self-administered urine testing only for use on several days around the expected middle of the menstrual cycle. The woman is advised to start testing usually about 3 days before the expected day of mid-cycle, which is estimated from her previous cycle lengths, and it is hoped that the anticipated surge of the LH hormone will be detected on one of the tested days. This expectation is not always fulfilled, which is one drawback of the method.

The "mid-cycle" is another inaccurate traditional concept. The middle of the menstrual cycle does not necessarily coincide with the day of ovulation, since many menstrual cycles are asymmetrical. On the whole, and particularly in baseline cycles of healthy women, the luteal phase after ovulation tends to be of constant length at 14 days +/- 1, whereas the duration of the pre-ovulation follicular phase varies from cycle to cycle and the menstrual cycle length varies with it. As discussed in more detail below, the variability of the follicular phase is due to the variable maturation rate of individual dominant follicles in different cycles. There are many instances where the luteal phase deviates from the expected constancy, as in abnormal cycles referred to as "short luteal phase" cycles. This is one of the problems that lead to difficulty to conceive when planned, and a useful tool of fertility diagnosis must be responsive to such problems.

The LH (luteinizing hormone) is monitored in the urine, into which it has cleared from the blood, hopefully appearing in the urine fairly soon, within about 4 to 6 hours. However, some much longer delays between the blood and urine LH peaks have also been reported. The hormone is released into the blood circulation from the pituitary gland in the woman's brain as a signal for the ovary to ovulate. The LH peak (or surge) is very narrow and therefore difficult to identify reliably because the time of testing for it may not coincide with the peak of the LH surge. For that reason, some LH kits aim to detect the onset of the LH surge (the initial increase from baseline) instead of the peak. The numerical definition of the onset varies from study to study and between kits. As a consequence, there is some inconsistency regarding the timing of ovulation after the detection of the LH hormone. Biological variability is also an inevitable factor in this inconsistency, adding to the methodological uncertainty factors.

Consider the inherent lack of accuracy associated with the onset of the urinary LH surge as an indicator of ovulation. The duration of the LH surge in blood is thought to be about 48 hours. Ovulation is thought to occur about 24 to 38 hours after the onset of the LH surge in blood, or about 9 to 16 hours after the peak of the LH surge. Therefore, ovulation occurs during the descending part of the surge. In winter and autumn months, women tend to ovulate in the evening or night hours since the LH peak tends to occur mostly in the morning hours [13]. A 12-hour shift occurs in the warm months of longer daylight. With the onset of the surge about 24 hours before the peak, there is a one-day uncertainty associated with the urine LH surge onset as a correlate of ovulation. This is because the lag may place ovulation either before or after midnight (24 to 38 hours after the onset of the blood LH surge and allowing at least 4 to 6 hours for the clearance into the urine). Note that this uncertainty also has a bearing on the accuracy of the Unipath urine analyzer ClearPlan Easy, discussed below.

Although the LH kits provide only an indirect indication of ovulation, the detection of the LH hormone is helpful to those experiencing difficulty to conceive. However, it is not a perfect tool, and it certainly cannot be used for birth control purposes, because the LH surge occurs too close to ovulation. As a consequence of the insufficiently long delay between the LH indication and ovulation, the LH kits miss a number of the fertile days before ovulation. Another drawback is that the kits have to be purchased every month, which would be too expensive if used over an extended period of time. It must also be noted that the occurrence of the LH surge does not guarantee ovulation, which only takes place if the LH signal from the brain is properly synchronized with the maturation of the ovum in the dominant ovarian follicle. Unlike the BioMeter, the hormone kits cannot recognize such important details of menstrual cycles.

One other advantage of the BioMeter over the LH kits is in that the monitor stores the menstrual profile data electronically. This is useful for possible later review by the physician attending to those women who seek help with infertility or other problems. The data can also be useful for those dealing with the projected timing of the upcoming birth. Menstrual cycle history is meaningful for the diagnosis and treatment of women's health problems, but only the basal body temperature charts have been available for the purpose. The BBT charts show whether or not the absence of the post-ovulation temperature rise may be indicative of some absent ovulations in the past. By comparison, the BioMeter data contain much more information useful to the physician, as will be shown below.

There are two other technologies that aim to achieve a more timely prediction of ovulation than the LH kits but these are not established as solidly in the marketplace. They are the ClearPlan Easy from Unipath Ltd. and the Cue fertility monitor from Zetek, Inc. Both these systems are more cumbersome to use and more expensive to produce than the BioMeter.

The ClearPlan Easy is an electronic colorimeter with a set of chemical reagents that, like the LH kits, analyzes the woman's urine but it analyzes for one other hormone in addition to the LH [14]. The instrument also improves the process over the kits in that it evaluates the indicator color changes electronically rather than relying on subjective judgement by the user. The second detected hormone is estrogen, which is known to peak within 14 to 24 hours before the LH peak in blood, which means about 24 to 36 hours before ovulation. In the regulatory mechanism of the menstrual cycle, estrogen stimulates the LH signal only when the estrogen levels in blood are sufficiently high for a sufficiently long time (14 to 27 hours).

Before the introduction of the ClearPlan Easy in the United States, Unipath Ltd. referred to their technology as a "personal contraceptive system". They hoped that the pre-ovulation rise of estrogen in the urine could provide an early enough predictive signal so as to indicate the beginning of the fertile period.

There has always been some doubt that such a delimitation of the beginning of the fertility window could occur sufficiently early and be reliable enough, even if they utilized the onset of the estrogen rise in a manner analogous to the use of the LH onset. The gradual increase of the E3G metabolite of estrogen in the urine, proceeding at different rates in different cycles, can hardly be associated with the beginning of the fertile period in a consistent, predictable manner.

The detection of the estrogen metabolite does not reflect the local interplay with progesterone and/or with the other timing mechanisms, but only reflects clearance of one of many metabolites of estrogen from peripheral blood circulation into the urine, after metabolic conversion in the liver. The stimulating effect of the ovarian estrogen on the brain hormone LH sets in too close to ovulation. In addition, the level of estrogen needed to stimulate the LH surge varies among women. This is of crucial significance because the inherent variability makes it difficult to set the triggering threshold with which to generate the predictive signal. By contrast, the BioMeter exhibits a dynamic range of readings that does not vary from person to person.

The ClearPlan Easy is subjected to some of the same problems as the LH kits. These problems include the uncertainty about when to expect the approach of the critical period and start testing, and the uncertain rate of hormone clearance from the blood into the urine. Another serious problem is the uncertainty associated with identifying the day of ovulation following the detection of the onset of the LH surge about 36 to 38 hours earlier. This assumed lag goes untested and inevitably introduces errors into the identification of the end of the fertile window. For, to cite from an authoritative review of the various timing mechanisms involved in the brain and in the reproductive tract, "it is a matter of conventional wisdom that severe perturbations in the external or internal environments can interrupt normal ovarian cyclicity in women" [15].

In view of the discussed problems, it is not surprising that the ClearPlan Easy is not referred to as a "personal contraceptive system" and that it is not recommended or sold for birth control use.

The fact remains that both hormonal markers tested for by the Unipath instrument are remote indicators of peripheral blood-borne stimulatory signals, with inherent uncertainties and without a detection of ovulation independent of the stimulatory signals. In addition, it is doubtful that those women who are not motivated by the desire to become pregnant would be willing to put up with the inconveniences of urine testing and with the recurrent monthly costs of the chemical reagents. We believe that the small tampon-like BioMeter, with its elegant feminine appearance, quick and easy procedure and affordable cost, is a more attractive proposition than the Unipath instrument.

There is one other potential competitor product, namely the Cue fertility monitor from Zetek, Inc. It is an electronic device with two probes. One probe is used in the mouth to measure the electrical resistance of the saliva, which provides an advanced warning of forthcoming ovulation. The other probe is for use in the vagina where it measures the electrical resistance of the vaginal fluids, in order to detect ovulation. The use of the Cue is quite complicated, and includes a fairly involved interrogation of the user by the instrument's data-processing electronics. The oral probe is applied daily from the beginning of the menstrual cycle, and later on the vaginal probe is applied as well, or instead, in order to confirm the ovulation event. A cumbersome user interface results from this dual sensor system, and this criticism is quite apart from the criticism of the fundamentals of this method of measuring the conductivity of the respective body fluids as a means of fertility monitoring.

Both Cue probes respond to the concentration of electrolytes, and particularly sodium, in the respective body fluids, saliva and vaginal mucus. Zetek's physiological rationale for the salivary resistance variations invokes the action of estrogen on the renin-angiotensin system, and the adrenal cortex increasing the secretion of aldosterone, which in turn alters sodium excretion and retention in body fluids. This is an indirect means of monitoring the reproductive cycle via the effects of the circulating pre-ovulatory estrogen on the liver and the kidneys. The consequence is a lack of precision. Zetek claims that the salivary signal occurs approximately six days before ovulation regardless of the length of the menstrual cycle. Below we shall see the consequence of this arbitrary constant on the performance of the Cue monitor in Zetek's own data.

The changes in the resistance of cervical mucus before and about the time of ovulation (due to higher sodium and water content) reflect changes in estrogen concentration in blood circulation. The estrogen concentration increases in preparation for ovulation as outlined above in relation to the Unipath instrument for urine analysis. The problem is that the cervical mucus changes are not sharply defined, and extend over several days. This was highlighted above in the discussion of the so-called peak mucus as a measure of the most fertile mucus in the subjective monitoring of mucus discharge according to the rules of natural family planning. The fertile mucus can be observed from 3 days before to 3 days after the estimated day of ovulation.

Another problem about the Zetek vaginal probe is that it samples, in fact, a mixture of vaginal fluids and cervical mucus, which further reduces the precision of the determination. This is a fundamental problem, not merely a small detail, because the two body fluids are very different from each other and serve two fundamentally different physiological roles.

Zetek's solution is a complex method of comparing the salivary and vaginal fluid resistance readings, in an effort to deduce the time of ovulation. Ovulation is assessed to occur on the day on which one of two conditions is satisfied. One condition is that a rise after a nadir in the vaginal resistance readings is observed - when such a day is within 6 days of the salivary resistance peak. The alternative condition is that ovulation is decided to have occurred exactly 6 days after the salivary peak - if no rise in the vaginal readings is obtained within that period. No rationale is offered for this arbitrary assignment.

Note that this postulate of 6 days for the estrogen-driven signal anticipating ovulation amounts to an implied postulate of a constant (subject- and cycle-independent) rate of maturation of the dominant follicle. It also implies the assumption that empirically the follicular phase should be constant and that it will be constant unless otherwise indicated by the vaginal resistance readings. These assumptions are wrong because they are not substantiated by reproductive physiology and endocrinology.

Both the oral and vaginal determinations by the Cue instrument are approximate, which is evidenced by the statistical distributions of the occurrences of ovulation as predicted (orally) and detected (vaginally) by the instrument. Zetek's 1989 patent [16] shows that, 6 days after the oral probe's peak of salivary resistance, the LH peak marker of ovulation was observed in only some 33% of menstrual cycles, with some 22% on day 5 and 25% on day 7. There is a bell-shaped statistical distribution around the sixth day after the predictive peak, and the distribution is 7 days wide.

This is a considerably wide spread, even if the inherent one-day uncertainty is allowed for the LH correlate of ovulation, which was used instead of the direct detection of ovulation by ultrasound scans of the ovaries. In over 150 menstrual cycles, there is a similar distribution around the Cue's vaginal marker of ovulation. The Cue ovulation marker coincides with the LH peak in only some 40% of menstrual cycles; about 33% of LH-estimated ovulations occurred one day later, about 15% two days later, and about 10% occurred one day before the Cue ovulation marker.

With such large margins of error, it is not surprising that the FDA has approved the Zetek instrument only for use by those seeking to get pregnant. The instrument is not approved for birth control use. This is somewhat unfortunate because the Vatican has expressed approval of such fertility monitors as the Cue. Perhaps even more important for the consideration of consumer appeal is the fact that the Cue procedure is rather involved. It can be argued that, even if it were FDA-approved for birth control use, the Cue would not be a very attractive proposition for those women who are not as motivated as those trying to get pregnant. For them, there are and/or will be other options including the simple, less expensive and more broadly usable BioMeter.

In this section, we reviewed the existing commercially available aids for fertility diagnosis and pointed out various aspects in which the BioMeter technology should improve upon those products. In closing this comparative overview, let us bring out the most important aspect of fertility status determination. It is that the technique must be responsive to the effects of a number of timing mechanisms involved in the menstrual cycle. Proper functioning of the menstrual cycle requires a proper interaction between the mechanisms of the various so-called clocks (e.g., a circamensual clock and a circhoral clock) and hormone generators (e.g., GnRH and LH pulse generators). It is known that various influences impinge on the functioning of the various components of the menstrual regulatory mechanism by disrupting the proper interaction between the brain and the reproductive tract.

Thus, disturbances in the menstrual cycle occur in response to exercise and physical demands, stress and emotional demands, and diet and nutritional demands [17]. As Michel J. Ferin writes, "with minimal reduction in (GnRH) pulse frequency, small undetected defects in the follicular maturation process may occur, whereas with a higher degree of pulse inhibition the follicular phase may be prolonged, and luteal phase deficiency, anovulation, and amenorrhea may develop." It will be shown below that the BioMeter technique appears to detect such phenomena as the different rates of follicular maturation in different menstrual cycles. This would seem to set the BioMeter apart from any of the reviewed techniques.



A brief overview of reports on the use of natural family planning or fertility awareness for birth control


This section is intended as a comment on fertility awareness/natural family planning (NFP) in different parts of the world. In 1984, a well-known proponent of NFP in England expressed [18] the hope that "new scientific techniques can be expected ... which will simplify the detection of the fertile phase and hopefully elucidate the grey area of the early fertile days, where most of the unplanned pregnancies occur". Flynn also commented that "many partners find the discipline enhances their sexual relationship and dialogue. Whatever the motivation, today more and more couples are happier to be in autonomous control of their bodies and their fertility."

In 1988, Kambic and coworkers at the Johns Hopkins University School of Hygiene and Public Health [19][20] observed that NFP is an important birth control method in certain countries, and that the success of the method depends on whether or not women can apply the method without teacher assistance. Those who could be self-sufficient had a 12-months continuation rate of 97% and a pregnancy rate of 1.7%. The authors also found that women who were at least 30 years old were significantly more likely to become self-sufficient in NFP, to use it without teacher assistance.

In the mid-nineties, the following two European groups published very encouraging results. The Dusseldorf, Germany NFP Study Group reported [21] on a 10 year prospective study and found distinct sociodemographic characteristics and behavioral and method use patterns, with overall low rates of unwanted pregnancies. In the study at Federation francophone pour le planning familial naturel in Brussels, Belgium [22], 59% of participants had a professional occupation, the average age was 32 years, and average fertility was 2 children per woman. No method failure was reported in 103 women-years of experience, with one user failure (one unintended pregnancy from unprotected intercourse during fertile phase). In the US, one investigation in Salt Lake City, UT found [23] that 43% of respondents were interested in learning more about NFP, while Kambic and Notare in Baltimore, MD concluded [24] that without greater commitment of resources, it is likely that NFP will continue to be a marginal method of family planning in the United States.

Yet, an Italian study in the late nineties concluded [25] that progress has been achieved in the teaching and application of NFP, and that "errors due to the method itself are few" whereas the main cause of any failure "seems to be the misapplication of NFP rules". The New Zealand Association of Natural Family Planning studied long-term continuation rates and found [26] that "the majority of subjects (>90%) were highly satisfied with NFP use, with the most common reasons for satisfaction being self-awareness, freedom from drugs, naturalness and effectiveness." This is consistent with the finding in Germany [27] of positive psychological aspects of NFP practice. Another German prospective long-term study established [28] that during "perfect use" the pregnancy rate at 12 months of use is 0.63%, and when only protected intercourse takes place during the fertile window, the pregnancy rate is 0.45%.

It would seem that NFP/fertility awareness could offer not only personal satisfaction but that it could have far-reaching societal consequences as well. The findings that the required discipline enhances the sexual relationship and dialogue [18], and that there is a reduction of "dominant" attitude in both men and women using NFP [27], suggest that this approach to birth control can contribute to the quality of marital partnerships. On a large scale, this could affect the divorce rates. Since more families in people's lifetimes can only mean more born children, a reversal of the trend due to improved quality of marital relationships could have the desirable impact on population growth trends, as well as on quality of life.



Related aspects of population dynamics and public health


The specialists concerned with population dynamics and public health have utilized a number of multi-center studies of natural family planning (NFP) as a source of data on reproductive safety issues and concerns such as pregnancy outcome. Unplanned pregnancies are a major safety concern in public health. It was concluded that "no increased risk of adverse pregnancy outcomes was observed among women who experienced an unplanned pregnancy while using natural family planning" [29]. In fact, the conclusion was that "women using NFP to plan their reproduction may be at particularly low risk" [30]. This contrasted with the complications associated with unplanned pregnancies (late pregnancy vaginal bleeding, vaginal infections, hypertension of pregnancy, proteinuria, glycosuria, and medication use) especially among the population that does not use NFP.

The issue of the sex ratio (males per 100 females born) appears unsettled in terms of the effects of such factors as the length of the follicular phase or the timing of insemination relative to the day of ovulation. However, the crucial problem here is again that these timing factors have been estimated rather than determined unequivocally [31]. The same problem interferes in animal studies, where the timing with respect to the onset of estrus (rather than ovulation) is simply not good enough and the data merely confuse the issue [32].

The Wilcox group found in 1995 that, among 133 births, conception cycles with short follicular phase tended to produce boys while those with long follicular phase produced more girls [33].

Several approaches and hypotheses of the sex ratio effects have appeared. Martin has proposed that the sex ratio depends on the length of the follicular phase because selection for Y spermatozoa decreases with improvements in cervical mucus permeability, which improves as the follicular phase lengthens. The sex ratio should indeed decline in long follicular phases [34]. It is important to note, for future study design and for data interpretation, that "debris left by earlier inseminations" interfere with this causality.

The epidemiological approach to sex ratio, coital rate, hormones and time of fertilization within the cycle by James [35] is a long-ongoing effort, and the hypothesis is that the sex ratio is causally dependent on parental hormone levels at the time of conception [36] [37]. James argues that "the regression of human sex ratio (number of males per 100 females at birth) on cycle day of insemination" is associated with "the regression of sex ratio on the duration of gestation (from last menstrual period to delivery)" [38].

It should be useful to analyze how these observed or hypothesized relationships may relate to each other. The overriding concern is the safety concern, and how these issues impinge on individual woman's or couple's feelings and experiences.



Conclusion


The fertility window problem was critically reviewed. It was argued that, in order to determine the width and the boundaries of the intermittently occurring period of female fertility, ovulation must not be merely estimated by means of indicators that do not unequivocally signify its occurrence. It was argued that the wide and skewed, rather than sharply defined, boundaries of the published fertile window profiles are probably in error due to the absence of definitive proof of ovulation in the respective studies.

It was shown that good evidence seems to exist for a three-day period of high probability of conception but that the position of the period with respect to ovulation is in doubt. It was pointed out that the real or practical fertility window may be wider than the at-most-three-day width expected from data on gamete lifetimes. However, the determination of the fertile window must include unequivocal confirmation of ovulation and not merely its estimation. With an unequivocal detection of the ovulation event, the timing of the fertile window should be equally unequivocal.

Two commercial electronic devices that claim being useful for personal fertility monitoring were critically reviewed in relation to the BioMeter. The fact remains that neither the Cue from Zetek Inc. nor the ClearPlan Easy from Unipath Ltd. have contributed to a definitive determination of the fertility window. These devices have not been approved by the FDA for birth control applications. It is the writer's contention that, for the determination of the fertility window, the Zetek and Unipath technologies work with inadequate indicators of fertility status. The inadequacy stems from their being indirect, remote indicators of hormonal input signals that stimulate ovulation as opposed to signifying the occurrence of ovulation.

Studies in a number of countries indicate that NFP and fertility awareness can be highly satisfactory birth control and family planning methods. This is so particularly for women at least 30 years old and those who can practice the methods autonomously, without teacher assistance. The BioMeter is meant to make that possible for many couples. We note that the same age group of people also has problems with impaired fertility, the solutions to which are also aided by technological tools for personal fertility monitoring.

It is proposed that a society-wide increase in the use of this approach to family planning could have far-reaching consequences: a) reducing the incidence of complications associated with unplanned pregnancies, and b) for divorce and population growth trends and for quality of life.



References


[1] Leon Speroff, Robert H. Glass and Nathan G. Kase, "Clinical Gynecologic Endocrinology and Infertility", Williams & Wilkins, 5th edition, 1994. First citation = page 238. Second citation = page 692. Third citation = page 219.

[2] Eli Y. Adashi, John A. Rock, and Zev Rosenwaks, editors, "Reproductive Endocrinology, Surgery, and Technology", Lippincott - Raven, 1996.

[3] John J. McCarthy, Jr. and H.E. Rockette, "Prediction of ovulation with basal body temperature", Journal of Reproductive Medicine 31 (No.8), Supplement, 742 - 747, 1986.

[4] R.F. Vollman, "The menstrual cycle", in E. Friedman, editor, Major Problems in Obstetrics and Gynecology, W.B. Saunders Co., 1977.

[5] A.E. Troelar, R.E. Boynton, G.B. Borghild and B.W. Brown, "Variation of the human menstrual cycle through reproductive life", International Journal of Fertility 12, 77, 1967.

[6] Thomas W. Hilgers, G.E. Abraham and D. Cavanaugh, "The peak symptom and the estimated time of ovulation", Obstetrics and Gynecology 52 (No. 5), 575 - 582, 1978.

[7] Thomas W. Hilgers, "The ovulation method of natural family planning", Pope Paul VI Institute Press, Omaha, 1992.

[8] Kamran S. Moghissi, "Cervical mucus changes and ovulation prediction and detection", Journal of Reproductive Medicine 31 (Number 8), Supplement, 748 - 753, 1986.

[9] Stephen L. Corson, guest editor, "Ovulation Prediction in the Treatment of Infertility. A Symposium", Journal of Reproductive Medicine 32 (Number 8), Supplement, 739, 1986.

[10] World Health Organization, "A prospective multi-centre trial of the ovulation method of natural family planning. V.", International Journal of Fertility 30 (3), 18 - 30, 1985.

[11] A.J. Wilcox, C.R. Weinberg and D.D. Berg, "Timing of sexual intercourse in relation to ovulation. Effects on the probability of conception, survival of the pregnancy, and sex of the baby", New England Journal of Medicine 333, 1517 - 1521, 1995.

[12] J.T. France, F.M. Graham, L. Gosling, P. Hair and B.S. Knox, "Characteristics of natural conception cycles occurring in a prospective study of sex preselection: fertility awareness symptoms, hormone levels, sperm survival, and pregnancy outcome", International Journal of Fertility 37 (4), 224 - 255, 1992.

[13] Machelle M. Seibel, "Luteinizing hormone and ovulation timing", Journal of Reproductive Medicine 31 (No.8), Supplement, 754 - 759, 1986.

[14] Michael Catt, John Coley and Paul J. Davis, "Monitoring method", U.S. Patent number 5,467,778, November 21, 1995.

[15] J. Hotchkiss and Ernst Knobil, "The hypothalamic pulse generator: The reproductive core", Chapter 7 in [2], pages 124 - 162.

[16] Ranjit S. Fernando and Jennine Regas, "Method for predicting optimum time for insemination", U.S. Patent number 4,836,216, June 6, 1989.

[17] Michel J. Ferin, "The menstrual cycle: An integrative view", Chapter 6 in [2], pages 103 - 121.

[18] Ann M. Flynn, "Natural methods of family planning", Clinical Obstetrics and Gynaecology 11 (3), 661 - 678, 1984.

[19] R.H. Gray and R.T. Kambic, "Epidemiological studies of natural family planning", Human Reproduction 3 (5), 693 - 698, 1988.

[20] R.T. Kambic and M.C. Martin, "Evaluating client autonomy in natural family planning", Advances in Contraception 4 (3), 221 - 231, 1988.

[21] C. Gnoth, P. Frank-Herrmann, G. Freundl, J. Kunert and E. Godehardt, "Sexual behavior of natural family planning users in Germany and its changes over time", Advances in Contraception 11 (2), 173 - 185, 1995.

[22] A. De Leizaola-Cordonnier, ""Natural family planning effectiveness in Belgium", Advances in Contraception 11 (2), 165 - 172, 1995.

[23] J.B. Stanford, J.C. Lemaire and A. Fox, "Interest in natural family planning among female family practice patients", Family Practice Research Journal 14 (3), 237 - 249, 1994.

[24] R.T. Kambic and T. Notare, "Roman Catholic Church-sponsored natural family planning services in the United States", Adv. Contracept. 10 (2), 85 - 92, 1994.

[25] M. Guida, G.A. Tommaselli, M. Pellicano, S. Palomba and C. Nappi, "An overview on the effectiveness of natural family planning", Gynecol. Endocrinol. 11 (3), 203 - 219, 1997.

[26] M. France, J. France and K. Townend, "Natural family planning in New Zealand: a study of continuation rates and characteristics of users", Adv. Contracept. 13 (2-3), 191 - 198, 1997.

[27] N. Klann, K. Hahlweg and G. Hank, "Psychological aspects of NFP practice", Internat. J. Fertility 33, Supplement, 65 - 69, 1988.

[28] P. Frank-Herrmann, G. Freundl, C. Gnoth, E. Godehardt, J. Kunert, S. Baur and U. Sottong, "Natural family planning with and without barrier method use in the fertile phase: efficacy in relation to sexual behavior. A German prospective long-term study", Adv. Contracept. 13 (2-3), 179 - 189, 1997.

[29] A. Bitto, R.H. Gray, J.L. Simpson, J.T. Queenan, R.T. Kambic, A. Perez, P. Mena, M. Barbato, C. Li, and V. Jennings, "Adverse outcomes of planned and unplanned pregnancies among users of natural family planning: a prospective study", Am. J. Public Health 87 (3), 338 - 343, 1997.

[30] P. Mena, A. Bitto, M. Barbato, A. Perez, R.H. Gray, J.L. Simpson, J.T. Queenan, R.T. Kambic, F. Pardo, W. Stevenson, G. Tagliabue, V. Jennings, and C. Li, " Pregnancy complications in natural family planning users", Adv. Contracept. 13 (2 - 3), 229 - 237, 1997.

[31] R.H. Gray, J.L. Simpson, A.C. Bitto, J.T. Queenan, C. Li, R.T. Kambic, A. Perez, P. Mena, M. Barbato, W. Stevenson, and V. Jennings, "Sex ratio associated with timing of insemination and length of the follicular phase in planned and unplanned pregnancies during use of natural family planning", Hum. Reprod. 13 (5), 1397 - 1400, 1998.

[32] R.W. Rorie, T.D. Lester, B.R. Lindsey, and R.W. McNew, "Effect of timing of artificial insemination on gender ratio in beef cattle", Theriogenology 52 (6), 1035 - 1041, 1999.

[33] C.R. Weinberg, D.D. Baird, and A.J. Wilcox, "The sex of the baby may be related to the length of the follicular phase in the conception cycle", Hum. Reprod. 10 (2), 304-307, 1995.

[34] J.F. Martin, "Length of the follicular phase, time of insemination, coital rate and the sex of offspring", Hum. Reprod. 12 (3), 611 - 616 , 1997.

[35] W.H. James, "Sex ratio, coital rate, hormones and time of fertilization within the cycle", Ann. Hum. Biol. 24 (5), 403 - 409, 1997.

[36] W.H. James, "The hypothesized hormonal control of human sex ratio at birth--an update", J. Theor. Biol. 143 (4), 555 - 564, 1990.

[37] W.H. James, "Parental hormone levels and the possibility of establishing that some mammalian sex ratio variation is adaptive", J. Theor. Biol. 140 (1), 39 - 40, 1989.

[38] W.H. James, "Cycle day of insemination, sex ratio of offspring and duration of gestation.", Ann. Hum. Biol. 21 (3), 263 - 266, 1994.

[39] W.H. James, "The status of the hypothesis that the human sex ratio at birth is associated with the cycle day of conception [letter]", Hum. Reprod. 14 (8), 2177 - 2178, 1999.