1 Biostatistics Branch, MD A3-03 and 2 Epidemiology Branch, National Institute of Environmental Health Sciences, P.O.Box 12233, Research Triangle Park, NC 27709, USA and 3 Dipartimento di Scienze Statistiche, University of Padua, Padua, Italy
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key words: Bayesian/fertile interval/ovulation/menstrual cycle/pregnancy probabilities
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Changes in semen characteristics with age have also been studied. The preponderance of the data suggest lower semen quality among men aged >50 years compared with men aged <30 years (Kidd et al., 2001), but there is limited evidence of declines with age in the 30s and 40s. Sperm motility is the parameter with the most evidence for an age-related decrease, even at relatively young ages. One study reports reduced post-thaw motility for men in their late 30s (Schwartz et al., 1983
). However, the data are inconsistent (Kidd et al., 2001
). For example, an analysis of 30000 IVF cycles for women with tubal sterility found no `important alteration of semen characteristic with age' (Guerin and deMouzon, 1997
). Genetic defects in gametes increase with age for both males and females (Martin and Rademaker, 1987
; Risch et al., 1987
; Hassold et al., 1996
). For females the rates clearly rise after age 35 years, but the evidence for men is less clear. In both the UK (British Andrology Society, 1999
) and the USA (American Society for Reproductive Medicine, 1998
) there are age limits of <40 years for semen donors in order to protect recipients from sperm that may be genetically defective.
The purpose of this study is to evaluate the effects of male and female age on natural fertility by carefully controlling for variation in sexual behaviour. We use data from a large multinational European prospective cohort study of couples practising natural family planning (Colombo and Masarotto, 2000). The probabilities of pregnancy associated with sexual intercourse on specific days relative to ovulation are estimated and compared across age groups. In addition, we investigate differences among age groups in length of the fertile window during which intercourse can result in a clinically detectable conception.
![]() |
Materials and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Briefly, ovulation days were estimated from the daily BBT data using published methods (Marshall, 1968; Colombo and Masarotto, 2000
). Although the last day of hypothermia prior to the post-ovulatory rise in basal body temperature clearly does not correspond perfectly with the release of the oocyte, previous data suggest that BBT-based estimates of ovulation day have a high probability of being within ±1 day of the true ovulation day (Dunson et al., 1999
). More accurate measures require assays of daily urine specimens or ultrasound monitoring, both of which are prohibitively expensive in large studies.
Bayesian statistical analysis approach
A BBT-based estimate of ovulation day was available for 5860 menstrual cycles from 770 women. Of these, 2539 cycles from 647 women had at least 1 day with intercourse reported in the 10 day interval beginning 7 days prior to and ending 2 days after the estimated ovulation day. This interval was chosen as a conservative first guess for the fertile window. Our analysis is based on the 433 detected pregnancies that occurred in these 2539 cycles, along with the daily intercourse records and estimated ovulation days.
Since the specific intercourse act responsible for a pregnancy cannot be determined with certainty in a menstrual cycle having multiple days with intercourse in a window of potential fertility, we followed the established approach of using a statistical model to estimate the day-specific pregnancy probabilities (Barrett and Marshall, 1969; Wilcox et al., 1995
, 1998
; Dunson et al., 1999
; Colombo and Masarotto, 2000
). Most previous estimates of the day specific probabilities have been based on a published model (Schwartz et al., 1980
), which assumes that batches of sperm introduced in the reproductive tract on different days mingle and then compete independently in attempting to fertilize the ovum. One of the primary drawbacks of the original Schwartz et al. model is that it implicitly assumes that all couples have the same probability of conceiving if they have intercourse at the same time relative to ovulation.
Extensions of the Schwartz et al. model have been developed for accommodating variability among couples in their fertility (Zhou et al., 1996; Dunson and Zhou, 2000
; Dunson, 2001
; Dunson et al., 2001a
,b
). In particular, the Bayesian approach of Dunson and Zhou (2000) accounts for known predictors of fertility (e.g. age) and unexplained heterogeneity among couples through a `cycle viability' factor that has a multiplicative effect on the day-specific pregnancy probabilities. An alternative Bayesian approach developed by Dunson (2001) allows couples to vary with respect to both the probability of pregnancy following intercourse on the most fertile day of the cycle and the decrease in the pregnancy rate on less fertile days. The latter approach is more flexible in that it allows differences in the width of the fertile interval that are distinct from differences in the pregnancy rate on the most fertile day. As we are interested in assessing the impact of male and female age on the duration of the fertile interval and the day-specific pregnancy probabilities within the fertile window, our analysis will be based on the Dunson (2001) model. Differences among age groups are tested by estimating posterior probabilities (PP), with PP
0.95 considered unlikely to be due to chance.
Since the ages of the male and female partners are highly correlated, we cannot simply include both male and female age in the model without facing problems with co-linearity. To avoid such problems, we instead included the age of the woman (categorized according to the intervals 1926, 2729, 3034 and 3539 years) and the difference in years of age between the male and female partners. There were 481, 923, 807 and 328 cycles and 103, 154, 140 and 36 pregnancies in the respective age categories. Most (76%) of the men were older than their partners, and the average difference in age between the men and the women was 2.4 years (SD = 3.5, interquartile range = 0.3, 4.4). Our hypothesis was that the effect of male age would be more pronounced among the older men, with minimal differences in fertility between 20 and 30 year old men. To assess this hypothesis, we estimated PP of an association between the age difference and the level and duration of fertility separately for men in different age categories.
To complete a Bayesian specification of our model, we chose mildly informative prior distributions for the baseline fertility parameters, based on a published analysis (Wilcox et al., 1998), and non-informative prior distributions for the age-effect parameters. We then used a Markov chain Monte Carlo algorithm (Gilks et al., 1996
) to obtain posterior summaries of each of the parameters.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Since it has been hypothesized that heterogeneity among couples in fecundity increases with age, we considered a model that allowed the magnitude of heterogeneity among couples in their fertility to vary depending on the age category. The resulting posterior means for the heterogeneity parameters were similar across age categories, with no evidence of an increasing or decreasing trend with age. Therefore, we simplified our analysis by focusing on a model with a single heterogeneity parameter.
The estimated pregnancy probabilities following intercourse on a given day relative to ovulation for average women aged 1926, 2729, 3034 and 3539 years (with their partners of the same age) are shown in Figure 1. The day-specific pregnancy probabilities decreased with age, with the pattern similar across age categories. To simplify assessment of differences between groups, we focused on a reduced model that incorporated age effects through a multiplicative cycle viability factor. Based on the simplified model, which produced estimates consistent with Figure 1
, women aged 2729 years were predicted to have lower pregnancy rates on average than women aged 1926 years given equivalent timing of intercourse (PP = 0.99). Women in the 2729 and 3034 year age categories had statistically indistinguishable rates, and there was evidence of a decline between the 3034 and 3539 year age groups (PP = 0.95).
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Data appropriate for estimation of the day-specific pregnancy probabilities have been collected in a British study of couples using natural family planning in the 1960s (Barrett and Marshall, 1969), in a North Carolina study of early pregnancy conducted in the 1980s (Wilcox et al., 1995
), and most recently in our multinational European study of daily fecundability (Colombo and Masarotto, 2000
). The data from the earlier studies have been analysed to estimate the fertile days of the menstrual cycle and the day-specific probabilities of conception (Royston, 1982
; Weinberg et al., 1994
; Royston and Ferreira, 1999
; Dunson and Weinberg, 2000
; Dunson et al., 2001a
). However, a detailed assessment of the effect of male and female age on the fertile window and daily probabilities of pregnancy has not been possible, given the limited number of participants and conceptions in earlier studies. Our data set is larger than prior studies in terms of numbers of couples, clinical pregnancies, and menstrual cycles with daily records of intercourse and menstrual bleeding. Moreover, many couples had intercourse on only one day in the fertile window since many were using fertility awareness methods to avoid conception. These factors allow more precise estimation of the day-specific probabilities of pregnancy. In addition, a recently developed statistical model provides the methodology for these analyses (Dunson, 2001
).
We find that the fertile interval lasts ~6 days and ends on the day of ovulation, in agreement with both the North Carolina and British cohorts (Dunson et al., 1999). The similarities to the results from the North Carolina study, which used a highly accurate surrogate for ovulation day based on urinary hormone metabolites (Baird et al., 1991
), suggest that bias caused by measurement error in our BBT-based marker of ovulation may be low. We had speculated that as men age, sperm survival after insemination would be reduced, thus reducing the length of the fertile window. The estimated fertile interval was indeed 1 day shorter for 40 year old men with 35 year old partners compared with 35 year old men with 35 year old partners. However, this difference was not statistically significant.
As expected, advancing female age was strongly associated with reduced fertility. The day-specific probabilities of pregnancy were observed to decline in women in their late 20s, slightly earlier than reported in the CECOS study of women with artificial insemination (Fédération CECOS, 1982). Nearly a 50% drop occurred between women in their early 20s and women in their late 30s. These estimates do not include the increased occurrence of spontaneous abortion that is evident in older women, but do include early, preclinical loss, which is not distinguishable from non-conception in these data.
Perhaps the most interesting result from our study is the observed decrease in fertility with male age, beginning in the late 30s. Studies of couples using assisted reproduction services show decreases in pregnancy rates with age starting as early as the 30s, but most are not statistically significant (Kidd et al., 2001). The limited data for non-clinical populations suggest reductions in fertility beginning in the 40s (Anderson, 1975
). Our analysis is much more convincing than prior studies because it accounts for female age and variability in sexual behaviour.
There are several possible biological mechanisms for this decrease. Achieving a clinical pregnancy depends upon testicular production of sperm that can mature properly, survive insemination and passage through the female reproductive tract and remain viable until the oocyte is available, undergo capacitation and the acrosome reaction, penetrate the zona pellucida, fertilize, and provide sufficiently normal genetic material for early development. Genetic defects in the form of sperm chromosomal abnormalities increase in frequency with male age, and the increase can be seen as early as the 30s (Risch et al., 1987). This could adversely affect sperm function and early embryonic development. Sperm motility may decline at these early ages (Schwartz et al., 1983
), but unless the male has quite low sperm counts, this alone is unlikely to directly affect fertility. Other semen characteristics are less predictably affected at such early ages (Schwartz et al., 1983
; Kidd et al., 2001
). Average FSH levels appear to increase in the 30s (Zumoff et al., 1982
), suggesting that age-related changes in the gonadalpituitary axis (Veldhuis, 1999
) may begin during midlife. In addition, the testes and prostate show morphological changes that might adversely affect both sperm production and the biochemical properties of the semen (Hermann et al., 2000
). Though few age-related differences in semen biochemistry have been reported (Schirren et al., 1977
), newly identified substances, such as fertilization-promoting peptide, that may be critical for sperm function (Fraser and Adeoya-Osiguwa, 2001
) have not been investigated. Whatever the mechanism, the reduction in fertility with male age in these healthy couples suggests that male gamete `overproduction' does not fully buffer against reproductive failure.
This study also documents the enormous heterogeneity in fertility among healthy couples that is not accounted for by age. The interquartile range in the probability of pregnancy on the peak day of the fertile window extends from a 20% probability of pregnancy to a 60% probability of pregnancy. Epidemiological studies have identified some of the factors associated with this variability in fertility, including prenatal exposures, sexually transmitted disease history, smoking, and occupational exposures (Bonde, 1999; Baird and Strassmann, 2000
), but much of this heterogeneity remains unexplained.
![]() |
Notes |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Anderson, B.A. (1975) Male age and fertility. Results from Ireland prior to 1911. Pop. Index., 41, 561567.[ISI]
Baird, D.D. and Strassmann, B.I. (2000) Women's fecundability and factors affecting it. In Goldman, M.B. and Hatch, M.C. (eds), Women's Health. Academic Press, New York, pp. 126137.
Baird, D.D, Weinberg, C.R., Wilcox, A.J. et al. (1991) Using the ratio of urinary oestrogen and progesterone metabolites to estimate day of ovulation. Statist. Med., 10, 255266.[ISI]
Barrett, J.C. and Marshall, J. (1969) The risk of conception on different days of the menstrual cycle. Pop. Stud., 23, 455461.[ISI]
Bonde, J.P. (1999) Occupational risk to male reproduction. Int. Arch. Occup. Environ. Health, 72, 135141.[ISI][Medline]
British Andrology Society (1999) British Andrology Society guidelines for the screening of semen donors for donor insemination. Hum. Reprod., 14, 18231826.
Colombo, B. and Masarotto, G. (2000) Daily fecundability: first results from a new data base. Demogr. Res., 3, 5.
Dunson, D.B. (2001) Bayesian models for distinguishing effects on the level and duration of fertility in the menstrual cycle. Biometrics, 57, 10671073.[ISI][Medline]
Dunson, D.B. and Weinberg, C.R. (2000) Modeling human fertility in the presence of measurement error. Biometrics, 56, 288292.[ISI][Medline]
Dunson, D.B. and Zhou, H. (2000) Bayesian modeling of fecundability and sterility. J. Am. Statist. Assist., 95, 10541062.
Dunson, D.B., Baird, D.D., Wilcox, A.J. and Weinberg, C.R. (1999) Day-specific probabilities of clinical pregnancy based on two studies with imperfect measures of ovulation. Hum. Reprod., 14, 18351839.
Dunson, D.B., Weinberg, C.R., Baird, D.D., Kesner, J.S. and Wilcox, A.J. (2001a) Assessing human fertility using several markers of ovulation. Statist. Med., 20, 965978.[ISI]
Dunson, D.B., Weinberg, C.R., Baird, D.D., Kesner, J.S. and Wilcox, A.J. (2001b) Modeling of multiple ovulation, fertilization and embryo survival. Biostatistics, 2, 131145.
Fédération CECOS, Schwartz, D. and Mayaux, M.J. (1982) Female fecundity as a function of age: results of artificial insemination in 2193 nulliparous women with azoospermic husbands. N. Engl. J. Med., 306, 404406.[ISI][Medline]
Ford, W.C.L., North, K., Taylor, H., Farrow, A., Hull, M.G.R., Golding, J. and ALSPAC Study Team (2000) Increasing paternal age is associated with delayed conception in a large population of fertile couples: evidence for declining fecundity in older men. Hum. Reprod., 15, 17031708.
Fraser, L.R. and Adeoya-Osiguwa, S.A. (2001) Fertilization promoting peptidea possible regulator of sperm function in vivo. Vitam. Horm., 63, 128.[ISI][Medline]
Gilks, W.R., Richardson, S. and Spiegelhalter, D.J. (eds) (1996) Markov Chain Monte Carlo in Practice. CRC Press, Boca Raton, FL.
Guerin, J.F. and deMouzon, J. (1997) Paternal age and fertility. Contracept. Fertil. Sex., 25, 515518.[Medline]
Hassold, T., Abruzzo, M., Adkins, K., Griffin, D., Merrell, M., Millie, E., Saker, D., Shen, J. and Zaragoza, M. (1996) Human aneuploidy: incidence, origin, and etiology. Environ. Mol. Mutagen., 28, 167175.[ISI][Medline]
Hermann, M., Untergasser, G., Rumpold, H. and Berger, P. (2000) Aging of the male reproductive system. Exp. Gerontol., 35, 12671279.[ISI][Medline]
Kidd, S.A., Eskenazi, B. and Wyrobek, A.J. (2001) Effects of male age on semen quality and fertility: a review of the literature. Fertil. Steril., 75, 237248.[ISI][Medline]
Marshall, J. (1968) A field trial of the basal-body temperature method of regulating births. Lancet, 2, 810.[ISI][Medline]
Martin, R.H. and Rademaker, A.W. (1987) The effect of age on the frequency of sperm chromosomal abnormalities in normal men. Am. J. Hum. Genet., 41, 484492.[ISI][Medline]
Risch, N., Reich, E.W., Wishnick, M.M. and McCarthy, J.G. (1987) Spontaneous mutation and perental age in humans. Am. J. Hum. Genet., 41, 218248.[ISI][Medline]
Royston, J.P. (1982) Basal body temperature, ovulation, and the risk of conception, with special reference to the lifetimes of sperm and egg. Biometrics, 38, 397406.[ISI][Medline]
Royston, J.P. and Ferreira, A. (1999) A new approach to modeling daily probabilities of conception. Biometrics, 55, 10051013.[ISI][Medline]
Sallmen, M. and Luukkonen, R. (2001) Is the observed association between increasing paternal age and delayed conception an artefact? Hum. Reprod., 16, 20272031
Schwartz, D., MacDonald, P.D.M., and Heuchel, V. (1980) Fecundability, coital frequency, and the viability of the ova. Pop. Stud., 34, 397400.[ISI]
Schwartz, D., Mayaux, M.J., Spira, A., Moscato, M.L., Jouannet, P., Czyglik, F. et al. (1983) Semen characteristics as a function of age in 833 fertile men. Fertil. Steril., 39, 530535.[ISI][Medline]
Schirren, C., Laudahn, G., Hartmann, E. and Heinze, I. (1977) Studies of the correlation of morphological and biochemical parameters in human ejaculate in various andrological diagnoses; 2nd report: biochemical parameters. Andrologia, 9, 95105.[ISI][Medline]
Veldhuis, J.D. (1999) Recent insights into neuroendocrine mechanisms of aging of the human male hypothalamicpituitarygonadal axis. J. Androl., 20, 1.
Weinstein, M. and Stark, M. (1994) Behavioral and biological determinants of fecundability. Ann. NY Acad. Sci., 709, 128144.[Abstract]
Weinberg, C.R., Gladen, B.C. and Wilcox, A.J. (1994) Models relating the timing of intercourse to the probability of conception and the sex of the baby. Biometrics, 50, 358367.[ISI][Medline]
Wilcox, A.J., Weinberg, C.R. and Baird, D.D. (1995) Timing of sexual intercourse in relation to ovulation: effects on the probability of conception, survival of the pregnancy and sex of the baby. N. Engl. J. Med., 333, 517521.
Wilcox, A.J., Weinberg, C.R. and Baird, D.D. (1998) Post-ovulatory ageing of the human oocyte and embryo failure. Hum. Reprod., 13, 394397.[ISI][Medline]
Zhou, H.B., Weinberg, C.R., Wilcox, A.J. and Baird, D.D. (1996) A random-effects model for cycle viability in fertility studies. J. Am. Statist. Assist., 91, 14131422.
Zumoff, B., Strain, G.W., Kream, J., O'Connor, J., Rosenfeld, R.S., Levin, J. and Fukushima, D.K. (1982) Age variation of the 24-hour mean plasma concentrations of androgens, estrogens, and gonadotropins in normal adult men. J. Clin. Endocrinol. Metab., 54, 534538.[Abstract]
Submitted on October 18, 2001; accepted on January 16, 2001.