Will it be a boy or a girl? A Harvard study shows that the odds of determining a baby's sex aren't always 50-50.

Gender reveal events have become a memorable moment in pregnancy, as couples, along with friends and family, gather to share the news. But what determines a baby's sex at the time of conception? While it is determined by the sperm's sex chromosomes in a process considered completely random, a new study indicates that other factors can tip the balance for a person or couple, determining whether their offspring is female or male.

The research, published in the journal Science Advances , included data from two large ongoing studies in the United States, the Nurses' Health Study II (NHSII) and the Nurses' Health Study 3 (NHS3) with birth records and genome-wide association data from 58,007 women . An analysis from which they hoped to answer questions such as whether the probability of having a son or a daughter is really a 50-50 coin toss for each birth or if there are genetic traits or reproductive factors that can be associated with a greater chance of having offspring of only one sex .

Regarding the results of this study, EL TIEMPO spoke with Dr. Jorge Chavarro , a Colombian scientist who is currently a professor of Nutrition and Epidemiology at the Harvard T.H. Chan School of Public Health and one of the authors of the research, which suggests that the probability of each family having sons or daughters is a 'loaded' coin toss, also influenced by maternal age and genetics .

The person who really initiated this idea was the second author, Dr. Bernard A. Rosner , a biostatistician who had seen several reports in the literature on this topic. They pointed out that when you look at aggregate data and consider family size, the sex distribution doesn't seem to be completely random, as you might expect. However, it's not entirely clear why this is the case. It's also a common observation: almost everyone knows at least one family in which all the children are boys or all the girls, which always raises the question of whether it's pure chance or not . So the main motivation was that curiosity: to understand how much of this is due to chance and how much is not.

We used data from two large studies: the Nurses' Health Study 2 and the Nurses' Health Study 3. From these, we analyzed the pregnancy history of the participants. What we found is that, although at the individual level—where each live birth is the unit of analysis—the sex distribution appears random, when we consider the presence of siblings (i.e., non-independent observations), the distribution deviates from what is expected. For example, in families with two children, a higher proportion of households have one boy and one girl than would be randomly expected. In larger families, with three, four, five, or six children, there are fewer cases with at least one boy and one girl than expected, and more cases of families with children of only one sex. The question then arises: Is this caused by biology or does it respond to patterns of reproductive behavior?

Other studies have shown that the decision to have more children is influenced by the sex of previous children. Specifically, in families with two boys or two girls, parents are more likely to seek a third pregnancy, unlike those who already have a boy and a girl. This pattern holds true even when controlling for family size. Since the advent of effective contraceptive methods in the 1960s, it is common for couples to decide to stop having children once they have one of each sex. For example, it is more common than expected to find families with boy-boy-girl or boy-boy-boy-girl, indicating a tendency to "stop" having the desired sex that was not present before.

In our data, we confirmed that this pattern exists. But beyond that, what we wanted to see is whether, independent of this reproductive behavioral factor, there are biological factors that could explain this deviation from chance-based expectations.

Yes. To study this, we eliminated the last child in each family from the analysis—since this is the one most likely to reflect a behavioral decision—and the associations became even stronger. This suggests that additional factors beyond reproductive behavior are at play. We then performed a diagnostic analysis of demographic, lifestyle, and reproductive factors of the women participating in the studies. We found that the woman's age at first birth was the only factor consistently associated with the likelihood of having children of only one sex.

We found that the older a woman was at the time of her first child, the greater the likelihood of having children of only one sex. This could be related to the fact that women who begin childbearing later tend to have smaller families and fewer opportunities to "balance" the sex ratio. But we also observed this association in families with two, three, or four children, suggesting that it is not explained solely by family size or reproductive lifespan. This likely points to other age-related processes that we cannot identify.

We know that there are many age-related changes in women's reproductive physiology that could play a role: changes in the hormonal patterns of the menstrual cycle, changes in vaginal pH, or in the length of the follicular cycle, for example. Any of these could be behind the signal we detected. But it's also possible that what we're interpreting as an effect of maternal age is actually reflecting something related to the father's age, since the two ages tend to be highly correlated in couples. That's a limitation of the study: we don't have information on the fathers, so we can't determine whether some associations are due to male or female aging.

We then performed a genome-wide association analysis (GWAS). We compared the entire genome of women who had children of only one sex (only boys or only girls) with that of women who had at least one child of each sex. We didn't see any signal associated with having only boys or only girls, but we did see a very clear genetic signal with the probability of having only boys and a completely different, also genome-wide, significant signal with the probability of having only girls. This suggests that there may be previously undescribed biological mechanisms associated with the specific survival of male and female embryos, but we don't know exactly why this is the case. What's quite intriguing is that the signal we see, for both boys and girls, is not related to genes that have anything to do with embryonic development, with the probability of having spontaneous abortions, or with the probability of infertility.

The signal we see in girls first is much stronger than the one we see with the probability of having only boys. Of the 25 strongest labels associated with girls, 20 are in the same gene, which has previously been associated with craniofacial development. But what does that have to do with the survival of female embryos? We have no idea, but it's absolutely fascinating. This is one of the most fun projects I've ever had in my life.

This raises many questions about sex determination at the time of conception and embryonic survival. Sex determination may be 50-50 in theoretical terms, but clearly that probability is not constant among all couples. When analyzed at the individual level, it appears random, but if we consider siblings, we see that some couples have a greater or lesser probability of having children of only one sex. And since that probability appears to be randomly distributed in the population, coupled with the common decision to "stop" after having both sexes, the observed pattern is accentuated.

There are two levels. The first is that it's simply fascinating from a scientific standpoint. The second is that it has practical implications. For example, for those planning their family: if you already have two girls, there's a higher probability that, if you have a third child, it will be another girl. It's not a certainty, but it's more likely (around 61 percent). So if you have another girl, at least you'll know it was statistically more likely and won't be surprised.

This study has important implications for reproductive health research. It clearly shows the need to consider correlations between pregnancy outcomes within the same woman or couple. It is known, for example, that the birth weights of two siblings are more correlated than those of two unrelated children. But these types of correlations are still frequently ignored in the medical literature. Our study is a clear example of how ignoring these correlations can lead to erroneous conclusions. For those new to research, it is also an excellent case study to illustrate the importance of considering correlated outcomes in family data analysis. This may not be a crucial topic for everyone, but it is definitely fascinating and, for those of us who work in research, tremendously enriching.