MATERNAL THYROID-STIMULATING HORMONE LEVEL IN THE FIRST TRIMESTER AND SEX RATIO AT BIRTH

[helvella]

I can't decide what to make of this abstract. As a simple observation, fine, it has some interest. But if people assume direct causality, they might try to manipulate TSH - to the detriment of themselves and offspring. (There are so many issues with gender selection, I won't start on that discussion.)

Endocr Pract. 2019 Apr;25(4):315-319. doi: 10.4158/EP-2018-0348.

MATERNAL THYROID-STIMULATING HORMONE LEVEL IN THE FIRST TRIMESTER AND SEX RATIO AT BIRTH.

Wang X, Sun X, Yang L, Tang R, Zhou J, Huang Y, Pan J, Chen X, Yang H, Chen Q, Chen Z, Mu L.

Abstract

Objective: Few studies have explored the influence of thyroid status on sex ratio at birth, and conclusions are inconsistent. The aim of this study was to determine if there is an association between serum thyroid-stimulating hormone (TSH) level in first trimester and sex ratio at birth.

Methods:

The study was a retrospective cohort study performed at a tertiary care center. From March 2014 to February 2017, a total of 4,822 women who had thyroid function testing during the first trimester were included. Study population was divided into five groups according to quintile of TSH level (≤0.60 mIU/L; 0.61 to 1.02 mIU/L; 1.03 to 1.44 mIU/L; 1.45 to 2.13 mIU/L; and ≥2.14 mIU/L). Logistic regression analysis was used to calculate the odds ratios (ORs) and 95% confidence intervals (CIs) of the percentage of male infants across the quintiles, with the lowest quintile as the reference category.

Results:

Median level of TSH was 1.27 mIU/L in women who delivered a boy, which was significantly higher than that in women who delivered a girl (1.15 mIU/L). After adjusting for age, gravidity, and parity, multivariate logistic analysis found that women in quintiles 3, 4, and 5 all showed significantly higher ORs for delivering a boy than those in quintile 1. In addition, after adjusting for age, gravidity, and parity, serum TSH was significantly associated with likelihood of having a boy (OR, 1.08; 95% CI, 1.03 to 1.13).

Conclusion:

Maternal TSH level in the first trimester is positively associated with the probability of delivering a male newborn.

Abbreviations:

CI = confidence interval; FT3 = free triiodothyronine; FT4 = free thyroxine; OR = odd ratio; SRB = sex ratio at birth; TBG = thyroxin-binding globulin; TSH = thyroid-stimulating hormone.

PMID: 30995430

DOI: 10.4158/EP-2018-0348

ncbi.nlm.nih.gov/pubmed/309...

[dtate2016]

It was true for me - I had a boy. Of course, in 1978 we did not know TSH levels nor measure them. But I’m quite sure I had untreated Hashimotos - it was very difficult to conceive even at 21 years old.

[Sybilla14]

Hmm... I thought the sex of the baby was determined at conception not during the 1st semester... Not convinced that tsh at 1st semester is necessarily correlated to that at conception. My tsh was 4.8 with the range top of 4.2 in January, conceived in early February but blood test in April showed tsh c.1 (albeit at 4pm and non-fasted). I had a girl.

[helvella in reply to Sybilla14]

I am not in the least convinced by the paper! Time of day at which blood tests were done can have at least as much effect on TSH as was observed - though we then have to ask why mothers of males had their blood drawn in the morning and of females in the morning!

But thyroid status at conception may affect which sperm achieves fertilisation. (With "first trimester" simply being around that time.)

[Sybilla14 in reply to helvella]

Indeed! I totally agree.

It’s the time of day, the effect of fasting or illness status that will all have an affect. I’d say, in the scheme of things, the difference in tsh between 1.27 and 1.15 is negligible even if ‘statistically significant’ using the statistical calculations applied on this study as it must vary in most people during the day or from day to day. And both tsh levels are comfortably within the normal/average ranges so assumed to be the same in effect.

[LAHs in reply to helvella]

That's just what I was thinking, 1.27 vs 1.15 mIU/L is within the diurnal variation I would think. It might even be within a range of the influenced of what everyone had for breakfast.

I think the maths was probably right on but the sampling technique and experimental control should have been spelled out a little more. (Perhaps it was in the source paper).

[jimh111]

A Chinese study, sometimes I wonder if they would benefit from re-introducing their 're-education' policy!!! The sex is determined by the male and the male foetus has more testosterone. Testosterone increases TSH. That's a possible explanation. It could be that the male being better just needs a little more TSH, or perhaps the female can get by with less attention. It's probably just a fluke, given the mass of the foetus is so small compared to the mother (no joke or insult intended here, just physics). On a more serious point it's worrying that such a study was approved.

[helvella in reply to jimh111]

Other than when the sperm has a clear M or F bias, is it true that the male determines gender? It seems entirely feasible to me that the female can influence whether M or F is more likely to reach the egg.

It would have been interesting to see what the males' thyroid statuses were prior to conception.

I wholly agree that the study is woeful. Even if the relationship between TSH and M/F is real and has some meaning, the mechanisms by which that comes about need to be elucidated. Is TSH in any way (directly or just as a marker) related to the M or F? Or, as you suggest, does the M or F of the foetus affect TSH?

[SilverAvocado in reply to helvella]

Yes, I was imagining something about the journey to the egg or the interaction of the sperm and the egg at the time they meet.

I believe it's a huge amount of activity and things that need to work out for the sperm that finally gets through, including entering the skin of the egg in the end, which is no simple matter.

[PiggySue]

Hmm...I have Hashi's and became quite ill after the delivery of first child -f, then went on to have m and f twins!

[diogenes Remembering]

It's just another example of probable coincidence with something else unknown and common to both measurements being the responsible factor. Too much of this a is linked to b and suggesting causality.

[SilverAvocado]

One notable thing about this abstract is it says the word 'significant' twice, but doesn't include a figure either time.

I can't remember the detail of odds ratios and their significance, but for comparing two lists of numbers (the TSHs for boy and girls babies) it will be a p-value, and the cutoff for a significant result is usually 0.05, meaning there a 1 in 20 probability that these results could just happen due to chance. If they've fudged things and mined the data in any ways that becomes even more probable.

Of course, as has been mentioned, if there's some other difference in mothers who are carrying boys that makes them push to get the morning appointment this is the result we'd expect

[LAHs in reply to SilverAvocado]

Yes, I agree. Usually statisticians are employed to grind the numbers and I got the impression that they seemed to know what they were doing. I think the data they were given was probably badly sampled, i.e. no strict scientific control was practiced in the collection of the data. When laymen can look at the results and point out that they defy common sense as many have done here, for me it is the margin of error for diurnal variation of TSH, then you have to go back to the drawing board and tighten up your observations.

Oh, and by the way, I could never grasp the concept of odds vs probability. Just when you understand the latter they throw in a bastardized version of it and call it odds. At least, that's how I saw it. If I put $5 on "Lilac" to win the Derby I realize we speak of the odds, but if we study form we think of the probability of it winning. Oh well, I guess I will have to Google it.

[LAHs in reply to LAHs]

Hi SiverA, here is a good explanation:

youtube.com/watch?v=Vu4x2DK...(Not working as expected?)

enjoy!

[SilverAvocado in reply to LAHs]

Hehe! Not sure 'enjoy' is quite the right word. I did understand these things sometime in the distant past. Fingers crossed it all comes back to me like magic when I watch the video!!

[SilverAvocado in reply to SilverAvocado]

Very easy to understand video! Confirmed my thought that a possible 1:1.08 (possibly as low as 1:03. A confidence interval, or CI in the abstract, is the range the true answer probably falls within) is a v small difference.

[SilverAvocado in reply to SilverAvocado]

Tried to see if I could use my new found knowledge from the video to guess at the real numbers in this abstract, and realised it is Odds Ratios, not Odds. I think that this means for every 1 unit change in the first variable (TSH) there is the greater chance of having a boy over a girl. With TSH this is a little confusing. Is the 1 unit a whole unit, the difference between a TSH of 1 and of 2. Because that's a huge difference, and the jump up to 3 is an even bigger deal, we are often interested in changes in the order of 0.1 for a healthy TSH.

This makes the finding seem even more ludicrous, as the vast majority of TSHs in the group will be in the healthy range (we hope), and therefore clustered in a band about 1unit wide. It really sounds like chance that maybe a few unlucky undiagnosed hypo women had boys.