Breast Milk and PFASs - Part 9

Continuing on from my last post...

In order for me to be correct in what I am about to show, my observation of the Breastfeeding Paper data must show that a geometric mean and IQR they used in their model is what I am seeing and was used to produce this statement:

The duration of exclusive breastfeeding was associated with increases of most PFAS concentrations by up to 30% per month, with lower increases during partial breast-feeding.

According to the Breastfeeding Paper's authors, the serum concentration for PFOS for the 80 mothers had a geometric mean of 6.0 ng/mL and an IQR of 5.2, 7.2.  This, if my understanding of an IQR is correct, means that 75% of the PFOS found in these 80 mothers was below 7.2 ng/mL.

Source: Breastfeeding Paper - Table 1 excerpt

Those 80 mothers were pulled from the 656 mothers with serum concentrations of PFOS collected at week 32 of pregnancy.

Here is where my angst is coming from.  Let's look at the data presented in the Antibody Paper for the 656 mothers from which those 80 mothers were pulled from:

Source: Antibody Paper

In particular, lets just look at the serum concentration for PFASs found in the 656 mothers at week 32 of pregnancy.

Source: Antibody Paper - Table 2 excerpt

Notice something?  The Antibody Paper from which the data came from tells us that the PFOS in the 656 mothers tested have a geometric mean of 27.3 ng/ml.

Something ain't right here.  How can these 80 mothers - pulled from the 656 mothers - have a PFOS serum concentration that far below the lower IQR of 23.2 ng/mL?

75% of the 80 mothers have a PFOS concentration in the lowest 25% of the 656 mothers?

That's...that's...highly unlikely!

In fact, looking at their data in Table 1, almost all of the four PFASs upper IQR value for the 80 women are below the lowest IQR value reported for the 656 women they came from.

That cannot be correct.  The odds of 80 mothers being so drastically different from the population they came from would, without benefit of a calculation, be exceptionally low, I contend that this serum data from the mothers is in error.

In contrast, the PFASs reported for the children at 11 months and 18 months fall very close to the IQR range reported for the mother's 32 week pregnancy concentration in the Antibody Paper.

Source: Breastfeeding Paper - Table 1 excerpt

At 5 years (60 months) the PFASs of these 80 kids align very close to the results report for the 656 in the Antibody Paper.

Source: Breastfeeding Paper - Table 1 excerpt

Source: Antibody Paper - Table 2 excerpt

I contend - and I could be wrong here - that the values they used in their prediction model for 0 month - the serum concentration of PFASs at 32 weeks of pregnancy - for these 80 mothers is incorrect.  I cannot accept the ng/mL values for 0 month that they report in Table 1.  I cannot accept it based on what is reported for the overall population for the mothers in Table 2 of the Antibody Paper which is where these 80 mothers came from.

If the glove don't fit....

So...the Breastfeeding Papers claims to show that:

The duration of exclusive breastfeeding was associated with increases of most PFAS concentrations by up to 30% per month, with lower increases during partial breast-feeding.

But for that to be true, the actual serum level of the child would need to be known at 0 month.  If you are going to assume that the serum level of the mother at 32 weeks is that value, then that value must be correct for your model to be sound.

Here is what I see - and why it has taken me a while to write these last posts.

I see that 328 women from the Antibody Paper had blood serum concentrations of PFOS between 23.2 and 33.1.  That's a range of 9.9 between 328 women.

According to the Breastfeeding Paper, 60 of the women - pulled from the Antibody Paper's 656 women - had PFOS concentrations 20.1 less than the lowest value seen in 492 (75%) of the women tested.

That cannot be correct.

Next post: Part 10