Why are boys more likely to be preterm than girls? Plus other related conundrums in human reproduction: Opinion

William H. James

The Galton Laboratory, University College London, Wolfson House, 4 Stephenson Way, London NW1 2HE, UK

Abstract

The present note offers two quite different points of interest. The first concerns the cause(s) of preterm labour. This is an important practical problem because of the associated morbidity and mortality. The second point of interest is theoretical. Some logically inter-related empirical propositions are formulated: none has been established but evidence for each is apparently evidence for all. However, as far as I know, there is no formal method of treating such material. It would be useful if methodologists would offer an opinion on whether the practical problem of the causes of preterm birth could be helpfully approached by the accumulation of data on the propositions outlined here. In my opinion, critics should not assess these propositions piecemeal, but should consider the whole edifice of propositions simultaneously. Obstetricians who confine their attention to the delivery of babies should extend it to the circumstances surrounding the conception of those babies. For there may lie the solution to the problem of why boys are more likely to be preterm than girls.

Key words: coital rate/duration of gestation/hormone concentration/sex ratio at birth/time of insemination within the cycle

Introduction

It is well known that the sex ratio (proportion male) of preterm infants is higher than that of infants born at term. In the US, among White singleton live births at <37 weeks gestation, ~55% are male, whereas among those born at term, ~50% are male (Cooperstock and Campbell, 1996Go; Cooperstock et al., 1998Go). What causes this excess of males among preterm births? The above authors suggest that somehow the sex of the fetus is associated with a mechanism that initiates labour. These authors acknowledge that `the precise biologic mechanism for the male effect reported here is unknown'. And they note that it appears not to be related to fetal weight because it apparently operates most powerfully during early gestation when fetuses are small, and not at all in later gestation when fetuses are large.

I suggest that the proposed mechanism does not exist, and that the regression of sex ratio on duration of gestation is a much-damped reflection (~9 months later) of a regression of sex ratio on cycle day of insemination within the cycle. In short, I suggest that boys are born earlier on the average because they are conceived earlier.

Since preterm birth is a major cause of morbidity, it is worth elaborating on the argument in detail. For this purpose, evidence will be reviewed that the sex of a human zygote is partially dependent on the time of insemination within the fruitful cycle. Evidence is of two sorts; direct and indirect.

Direct data

There are five non-human mammalian species for which it has been reported that the time of insemination within the cycle is directly associated with the sex of the resulting offspring, i.e. white-tailed deer, Barbary macaque, golden hamster and Norway rat (for references, see James, 1996) and mouse (Tarín et al., 1999Go). With regard to human beings, I conclude from previous data that there is a regression of sex ratio on cycle day of insemination such that boys are conceived earlier than girls (James, 1971Go). More recently, Guerrero (1974) and Harlap (1979) elaborated on this, offering evidence that the regression is U-shaped (rather than monotonic declining), and a meta-analysis concurred (Gray, 1991Go).

However, since the publication of this last paper, work by the same author and others has failed to offer much direct support to the hypothesis. So in a further meta-analysis (James, 2000Go), I considered the data reproduced in Gray (1991) and the four further sets of data known to me (making 10 data sets in all). Each of these sets may be cast in the form of a 2x2 frequency table with sex being one variable (male versus female) and cycle day of insemination (most fertile versus other days) the other variable. In summarizing the result of a number of such data sets, one may use the Mantel–Haenszel method. This is most comprehensibly described in Snedecor and Cochran (1967). The overall Mantel–Haenszel test statistic, z (a normalized deviate), took the value 2.87 (P < 0.005). Thus the result is highly significant, though it may be acknowledged that one's interpretation of it is dependent on one's evaluation of the raw data, especially those of Guerrero (1974) and Harlap (1979). Both of these papers were followed by a flurry of critical correspondence in the New England Journal of Medicine. In my opinion, this correspondence identified flaws in methodology, but not flaws that might be responsible for the reported overall U-shaped regression.

In this context, it is worth noting that one group of workers (Weinberg et al., 1995Go; Wilcox et al., 1995Go) found no support for the present suggestion. Instead, they presented data linking the sex of zygote with the length of the follicular phase, i.e. the interval between the last menstrual period (LMP) and ovulation, rather than with the interval between insemination and ovulation. However, as these authors acknowledge, their sample was comparatively small; moreover as I pointed out (James, 1997aGo) their suggestion would seem to imply variation of offspring sex ratio with mean maternal cycle length – evidence for which seems absent. Lastly, Fanchin et al. (1998) noted that when human chorionic gonadotrophin (HCG) is administered to women, the length of the follicular phase apparently is not correlated with the sex of the resulting zygote.

At first sight, it might seem that we have reached the limits of useful speculation here. But I now wish to urge that a wealth of indirect data support the hypothesis.

Indirect data

If x correlates with y, and y correlates with z, then (unless countervailing factors operate), x correlates with z. In such a case, Occam's Razor suggests that the possibility of such countervailing factors be provisionally ignored. The following considerations would constitute powerful evidence, unless some hitherto unidentified phenomenon were shown to be operating. There are a number of related propositions each of which may be designated by a capital letter.

  1. The sex of a zygote is associated with parental hormone concentrations around the time of conception, high concentrations of oestrogen and testosterone being associated with the subsequent births of boys, and high concentrations of progesterone and gonadotrophins with girls.
  2. There is a U-shaped regression of human offspring sex ratio (proportion male) on time of formation of the zygote within the fruitful cycle.
  3. There is a positive association between parental coital rate around the time of conception and offspring sex ratio.
  4. There is a-shaped regression of human sex ratio on duration of gestation (reported LMP to delivery).
  5. In some species of polytocous mammals, the variances of the distributions of the combinations of the sexes within litters is sub-binomial, i.e. there are more litters with equal numbers of males and females than would be expected under binomial sampling.

Using Occam's Razor in the sense above, if A were true, then B must be true. And if B were true, then C, D and E must all be true.

In the rest of this note, it will be shown how proposition B is dependent on A, and how C, D and E may be derived from B. The strength of the relevant direct data on all five propositions will be indicated.

Proposition A
I have shown that under a very wide range of circumstances, the hormone levels of both mammalian parents around the time of conception are associated with the sexes of the resulting offspring (James, 1996Go, 1999Go). Some of these data suggest that the relationship is causal. The overall suggestion is that high levels of parental oestrogen and testosterone are associated with subsequent births of boys, and high levels of progesterone and gonadotrophins with girls. The mechanism underlying this phenomenon is not relevant in the present context, though I have suggested one (James, 1997bGo). I would suggest that this is a hypothesis for (parts of) which there is powerful evidence.

Proposition B
The direct data on this proposition have been described above. It is implied by proposition A in that the mother's gonadotrophin/oestrogen ratio rises and then falls across the fertile interval, thus ex hypothesi being responsible for a U-shaped regression of offspring sex ratio on cycle day of conception. In other words, if A were true, then B would be expected to be true. The direct evidence for this hypothesis is much thinner than that for A.

Proposition C
Direct data relating coital rates to human sex ratios are not strong (James, 1997cGo). However, such a relationship has been reported in other mammalian species, e.g. horse, rabbit, rat, mouse, cattle and possibly some species of seal (for references, see James, 1996). Roberts (1978) showed mathematically how the time of fruitful insemination within the cycle may be expected to vary with coital rate. In short, if B were true, C would be expected to be true too. The direct evidence for this hypothesis is weak in man.

Proposition D
As noted above, sex ratio reportedly declines with duration of gestation, boys comprising ~55% of preterm births (<37 weeks gestation) but only ~50% of term births (Cooperstock and Campbell, 1996Go; Cooperstock et al., 1998Go). However these authors do not notice (let alone explain) that after 42 weeks reported gestation, the sex ratio rises again. This latter rise occurred in respect of singleton US White live births for all 11 years, 1966–1976 inclusive (James, 1994Go); and in respect of three large English samples (Karn and Penrose, 1951Go; McKeown and MacMahon, 1956Go; Milner and Richards, 1974Go). Thus proposition D is on a much firmer empirical basis than the other propositions considered here. Moreover, I demonstrated that (subject to an assumption stated below) the magnitude and shape of the regression of sex ratio on cycle day of insemination reported by Guerrero (1974) closely predicted the magnitude and shape of the regression of sex ratio on duration of gestation reported in the US Vital Statistics (James, 1994Go). This phenomenon is present every year in the US Vital Statistics and suggests that the interval is ~1 day shorter for boys than girls. The assumption underlying this demonstration was that the distribution of `true' gestation (conception to delivery) is not affected by time of fertilization within the cycle nor by the sex of offspring. Lastly, it may be noted that for a given reported duration of gestation, the variance of birth weight is greater for males than for females (Milner and Richards, 1974Go). This rules out the possibility that reporting bias (mistaken dates) was wholly responsible for the increase in sex ratio of births with gestations reported at >42 weeks. (The greater male variance in birth weight apparently follows a really greater male variance in conception day).

It is clear that if B were true, then D would be expected to be true in some sense. In fact though, the magnitude of the effect at delivery that could be caused by events at conception is (to me) counter-intuitive. My demonstration of this relationship (James, 1994Go) must throw doubt on Cooperstock's proposed sex-related mechanism for initiating preterm delivery. This is so because the above reasoning would explain the whole of the regression of sex ratio on duration of gestation, whereas Cooperstock's proposal could only explain the decline across 32–42 weeks, and not the subsequent rise.

Lastly, cross-species evidence adduced by McKeown and MacMahon (1956) throw doubt on the notion that fetal weight (or a correlate of it) initiates labour, because (i) in the guinea-pig there is a substantial decline of sex ratio with duration of gestation (as in man). Yet there is little difference between the weights of pups of the two sexes of this species. And (ii) in the cow, males (which are heavier) are born later. It would be interesting to learn whether they are conceived later too. Rorie (1999) has recently reviewed this topic. Unfortunately no consensus has been reached, perhaps because of the multiplicity of variables which potentially affect offspring sex ratio (artificial versus natural insemination; chemical induction of ovulation or oestrus; mode of detecting oestrus, etc). Overall, I regard proposition D to be so strongly supported as to be established, although in need of further interpretation.

Proposition E
The main application of this proposition is to litters of non-human mammals. In principle it may apply to human dizygotic twins and trizygotic triplets: but in practice (though the point has been intensively studied) the sex distributions within these categories are not finally established (James, 1992Go). However, though the point requires clarification in human beings, it is well established in a number of other species. Sub-binomial variance has been reported in litters of pigs, rabbits, mice, sheep, golden hamsters and Wistar rats (for references, see James, 1996).

The relevance of this is as follows. If proposition B were true, then since ex hypothesi zygotes are formed across time within the fertile interval, P (the probability of a male zygote) would vary from zygote to zygote within a litter. This would constitute an example of Poisson variation : and it is a standard result in probability theory that Poisson variation is associated with sub-binomial variance (Edwards, 1960Go). In short, if proposition B were true, then proposition E would also be expected to be true. The credibility status of this proposition is similar to that of proposition D.

Comment

I know of no formal method of treating a number of interdependent uncertain empirical propositions such as are outlined above. Evidence for any of them seems (in the absence of countervailing data) evidence for all. In short, propositions A, C, D and E have been considered here because of the (admittedly unquantified) extra weight they confer on B. Indeed, I should welcome guidance in the assessment of the additional weight they do confer on B.

When considered in this light, the evidence for B seems formidable. And if this proposition were true, then the regression of sex ratio on duration of gestation is explained without invoking a sex-related factor that initiates labour. So I suggest that boys are born ~1 day earlier than girls because they are conceived ~1 day earlier. If this were correct, the excess preterm boys throw no light on the mechanisms initiating labour.

Further research

Since it is important to elucidate these mechanisms, further study should be made of proposition B (that there is variation in offspring sex ratio by time of formation of the zygote within the cycle). This proposition (which might otherwise seem a curiosity) is important in this context.

Without laboratory resources and the collaboration of mothers-to-be, the exact time of initiation of a given pregnancy is not known. Hitherto, calculations have been based on the supposition that pregnancy starts ~14 days after the first day of LMP. However this is only a rough approximation and Larsen et al. (2000) have shown that a superior estimate may be based on the fetal biparietal diameter according to the formula of Persson and Weldner (1986). Unfortunately, male fetuses grow faster than females, so I suggest that Persson and Weldner's formula be adjusted to yield sex-specific estimates of gestational age. With such a resource, it might be possible to establish whether boys really are conceived earlier than girls on the average.

Notes

This opinion was previously published on Webtrack, July 3, 2000

References

Cooperstock, M.S., Bakewell, J., Herman, A. and Schramm, W.F. (1998) Effects of fetal sex on risk of very preterm birth in twins. Am. J. Obstet. Gynecol., 179, 762–765.[ISI][Medline]

Cooperstock, M. and Campbell. J. (1996) Excess males in preterm birth : interactions with gestational age, race and multiple birth. Obstet. Gynecol., 88, 189–193.[Abstract/Free Full Text]

Edwards, A.W.F. (1960) The meaning of binomial distribution. Nature, 186, 1074.[ISI][Medline]

Fanchin, R., Righini, C., Olivennes, F. et al. (1998) Female and male human embryo growth rates are similar before the eight cell stage. Am. J. Obstet. Gynecol., 178, 45–49.[ISI][Medline]

Gray, R.H. (1991) Natural family planning and sex selection : fact or fiction. Am. J. Obstet. Gynecol., 165, 1982–1984.[ISI][Medline]

Guerrero, R. (1974) Association of the type and time of insemination within the menstrual cycle and the human sex ratio at birth. N. Engl J. Med., 291, 1056–1059.[ISI][Medline]

Harlap, S. (1979) Gender of infants conceived on different days of the menstrual cycle. N. Engl J. Med., 300, 1445–1448.[Abstract]

James, W.H. (1971) Cycle day of insemination, coital rate and sex ratio. Lancet, i, 112–114.

James, W.H. (1992) The current status of Weinberg's Differential Rule. Acta Genet. Med. Gemellol., 41, 33–42.[ISI][Medline]

James, W.H. (1994) Cycle day of insemination, sex ratio of offspring and duration of gestation. Ann. Hum. Biol., 21, 263–266.[ISI][Medline]

James, W.H. (1996) Evidence that mammalian sex ratios at birth are partially controlled by parental hormone levels at the time of conception. J. Theor. Biol., 180, 271–286.[ISI][Medline]

James, W.H. (1997a) Follicular phase length, time of insemination, mean cycle length, season of mother's birth and sex ratio of offspring. Hum. Reprod., 12, 398.[ISI][Medline]

James, W.H. (1997b) A potential mechanism for sex ratio variation in mammals. J. Theor. Biol., 189, 253–255.[ISI][Medline]

James, W.H. (1997c) Coital rates and sex ratios. Hum. Reprod., 12, 2083–2084.[ISI][Medline]

James, W.H. (1999) Further evidence relating offspring sex ratios to parental hormone levels around the time of conception. J. Theor. Biol., 197, 261–263.[ISI][Medline]

James, W.H. (2000) Analysing data on the sex ratio of human births by cycle day of conception. Hum. Reprod., 15, 1206–1207.[Free Full Text]

Karn, M.N. and Penrose, L.S. (1951) Birth weight and gestation time in relation to maternal age, parity and infant survival. Ann. Eugen., Lond., 16, 147–164.

Larsen, T., Nguyen, T.H., Greisen, G. et al. (2000) Does a discrepancy between gestational age determined by biparietal diameter and last menstrual period sometimes signify early intrauterine growth retardation? Br. J. Obstet. Gynaecol., 107, 238-[ISI]

McKeown, T. and MacMahon, B. (1956) Sex differences in length of gestation in mammals. J. Endocrinol., 13, 309–318.[ISI][Medline]

Milner, R.G.D. and Richards, B. (1974) An analysis of birth weight by gestational age of infants born in England & Wales, 1967–71. J. Obstet. Gynaecol. Brit. Comm., 81, 956–967.

Persson, P.H. and Weldner, B.M. (1986) Normal range growth curves for fetal biparietal diameter, occipito frontal diameter, mean abdominal diameters and femur length. Acta Obstet. Gynecol. Scand., 65, 759–761.[ISI][Medline]

Roberts, A.M. (1978) The origins of fluctuations in the human secondary sex ratio. J. Biosoc. Sci., 10, 169–192.[ISI][Medline]

Rorie, R.W.(1999) Effect of timing of artificial insemination on sex ratio. Theriogenology, 52, 1273–1280.[ISI][Medline]

Snedecor, G.W. and Cochran, W.G. (1967) Statistical Methods. 6th edn. Iowa State University Press, Ames, IO, USA, pp. 256.

Tarín, J.J., Perez-Albala, S., Aguilar, A. et al. (1999) Long term effects of postovulatory aging of mouse oocytes on offspring: a two generational study. Biol. Reprod., 61, 1347–1355.[Abstract/Free Full Text]

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

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, 1517–1521.[Abstract/Free Full Text]