I made it quite clear that I did not think the paper was very good. Yes, it included data on tooth decay and some blood and tissue concentrations but very few conclusions can be drawn form them – and the authors were careful not to do so but did not do further testing to enable conclusions.When I have attempted to published unfinished research like that reviewers have told me to go back and do the experiments – which I have.

You say “I don’t think you can get more definitive than verifying side effects with placebo testing. There is certainly no “speculation” as you put it.” But there was absolutely no data reported for side effects – the statement was purely speculative (and I can only think based on their reading of the literature – they certainly did not present data).

You say:

“Do you think reported side effects should be investigated and how do you propose to do that?” – Yes of course. I myself don’t propose to do anything – its not my job. But researcher have been doing it. You might not like the results because they do not confirm you bias, but the research still exists.

“The method Feltman and Kosel used is one possible option. What do you think?” No they did not use an option to look at side effects. They did not include any methodology to do so or any results. Absolutely no data.

“How would you design side effect testing Ken?” I would design experiments to produce objective data. its not hard to do for a genuine researcher. I, for example, discuss one piece of published research on fluoridation side effects in my article “Fluoride sensitivity – all in the mind?” (https://openparachute.wordpress.com/2013/08/18/fluoride-sensitivity-all-in-the-mind/)

I take this as an indication that not only have you backed away from claiming flaws” in the Dunedin research, its an admission you actually don’t understand that and similar research. You are not capable of considering the actual data and you simply search for statements from “authority” figures. Even someone like Peter who has clearly not even looked at the data.

At least that pro-fluoride fanatic has got some sense of honesty, even if it was only on that point.

Do you think reported side effects should be investigated and how do you propose to do that?
The method Feltman and Kosel used is one possible option. What do you think?
How would you design side effect testing Ken?

“All criticisms that have been levelled at the Green study, can be levelled at the Broadbent study as well as a whole lot more. I was going to point them all out but, really, it is obvious to anyone who is willing to look at this honestly.”

I am disappointed you refuse to “point out” the criticisms you have of these two papers. Progress in science is driven by such critiques and always look forward to them.

I have discussed above “limitations” with the Broadbent paper – simple detection sensitivity which Broadbent himself acknowledges. (The study of Aggeborn & Öhman (2016) has far more sensitivity than any so far reported studies in this area because of the much larger sample size used and but still could not show a IQ effect) But I am far more concerned with the limitations in the Green et al paper because of its refusal to supply important statistical analysis results.

For example:

1: Their Table 1 lists the IQ of children for mothers from fluoridated and unfluoridated areas. The values are not different, statistically. Useful information – but the authors do not comment on that data in their discussion at all. I think that is dishonest.

2: Questions have been raised about the way the authors subdivided the data after the event. However, I am also concerned that the data for boys is not normally distributed so does not really satisfy the requirements for the linear regression used. This is perhaps a qualitative judgment, but one they hint at in their descriptions. They decided not to transform the data – perhaps they should have but I suspect this would have revealed there was no statistically significant relationship.

3: More importantly the relationship they report for boys is extremely weak – explaining only 1.3% of the data. It has absolutely no predictive power so their talk of a specific IQ drop with an increase in MUF is completely meaningless – it refers only to the best-fit line explaining a tiny proportion of the data and is not relevant to the data as a whole. This is immediately obvious to anyone who actually looks at the data presented in their figures.

4: I think it is dishonest for authors to report only meaningless p-values from their statistical analyses and not report the other important factors like the R-squared value. I can understand that the authors may not have wished to bring attention to how meaningless their relationship was but their omission should not have been allowed by the journal or reviewers (perhaps this gives us an idea of what the arguments in the editorial board about this paper involved).

5: I say dishonest rather than neglectful because these authors knew what they were doing. Till was involved with this paper and the Malin & Till (2015) ADHD paper. In that paper they had no hesitancy of reporting their relationships explained 22 – 34% of the ADHD variance (they reported R-squared values) but refused to report the same figures in the Green paper. Presumably because an R-squared value indicating only 1.3% explanation of IQ variance would expose how weak the relationship really is.

The paper reports data on fluoride levels in blood and placental tissue and cord (Figs I & II, Table 1, 2 & 3) and dental health of the children (tooth decay data in Tables 4A, 4B and 5. They also included data on dental health from other studies.

Their summary includes the comments:

“Data is presented showing that fluoride ingested by gravid women enters the maternal circulation, is stored in the placenta and passes through the placental barrier to enter the fetal blood supply.
Evidence is presented that fluoride in the fetal blood supply affects the developing teeth to make them more resistant to dental caries.
No unusual effects have been observed on the blood pictures of children who ingested fluoride.”

There is no data on the speculated side effects so the paper is of no value in any discussion of such side effects. It says a lot when anti-fluoride activists have to use a speculative statement completely devoid of any data to support their claims. (While at the same time being silent about the paper reporting beneficial effects of fluoride supplements on dental health).

Gus, this is a poor quality paper (although we should perhaps make allowance for its age). it is not readily available either. I am surprised you cannot find anything more substantial or modern to support you claims.

“One percent of our cases reacted adversely to the fluoride. By the use of placebos, it was definitely established that the fluoride and not the binder was the causative agent. These reactions, occurring in gravid women and in children of all ages in the study group affected the dermatologic, gastro-intestinal and neurological systems. Eczema, atopic dermatitis, urticaria, epigastric distress, emesis, and headache have all occurred with the use of fluoride and disappeared upon the use of placebo tablets, only to recur when the fluoride tablet was, unknowingly to the patient, given again. When adverse reactions occur, the therapy can be readily discontinued and the patient or parent advised of the fact that sensitivity exists and the element is to be avoided as much as possible.”
SOURCE: Feltman R, Kosel G. (1961). Prenatal and postnatal ingestion of fluorides – Fourteen years of investigation – Final report. Journal of Dental Medicine 16: 190-99.

So, the chance of actual IQ differences being outside those ranges is only 5%.

Broadbent, like all scientists, will acknowledge the limitations imposed by sample numbers etc., on the ability to detect such small differences. But Aggeborn & Öhman (2016) themselves could not find an IQ difference due to fluoride and reported confidence intervals of “(-1.8560–0.5546) for the specifications without covariates or fixed effects and (-0.1776–1.0311) for the specifications with all covariates and fixed effects, when fluoride is increased by 1 mg/l.”

So you and Connett are getting pretty desperate to continue arguing that the studies (Broadbent et al., Green et al., and Aggeborn & Öhman) simply lack the statistical power to detect the minuscule IQ differences you happen to believe in.

I should note that many of these studies do have the statistical power to detect other differences like effects on dental health and quality of life. Aggeborn & Öhman (2016), for example, report in their abstract:

“Taking all together, we investigate and confirm the long-established positive relationship between fluoride and dental health. Second, we find precisely estimated zero-effects on cognitive ability, non-cognitive ability and math test scores for fluoride levels in Swedish drinking water. Third, we find that fluoride improves later labor market outcomes, which indicates that good dental health is a positive factor on the labor market.”