Connett & Hirzy do a shonky risk assesment for fluoride

Paul Connett, executive director of the Fluoridation Action Network (FAN), told me, during our fluoride debate, that he was writing a scientific paper defining a lower safety limit for fluoride than currently accepted. Nothing has been published yet – although a recent FAN newsletter did refer to a risk assessment paper by him and Bill Hirzy currently under review. I look forward to reading this paper, but I am not holding my breath as neither author has an impressive publication record.

Connett described his risk assessment for fluoride in the debate (see Fluoride debate: Paul Connett’s Closing statement) and he and Hirzy have also made comments on this lately. They are rejecting the current risk assessment, based on the incidence of severe dental fluorosis, and using the incidence of IQ deficits instead. To this end, they are heavily promoting the work of Choi et al., (2012) and Xiang et al., (2003) (which reported IQ deficits in areas where fluorosis is endemic). They are also attempting to rubbish published research (such as Broadbent et al., 2014) which show no significant IQ deficits at fluoride concentrations used in community water fluoridation.

Connett and Hirzy have also organised campaigns to congressional representatives in their effort to force a downward revision of the Environmental Protection Agency’s (EPA) standards for fluoridation.

Connett’s approach is a desk study – these guys are not going  to dirty their hands by doing their own research to get useful data. They are taking a value which they claim represent the lowest concentration of fluoride in drinking water below which no IQ deficit was found. They then apply “safety factors” to effectively conclude the only safe concentration is zero (see Scientist says EPA safe water fluoride levels must be zero)!

I will be a bit surprised if they manage to squeeze their paper though a decent review process because their approach is shonky. Look at the way they use the data from Xiang et al., (2003). (I have used the presentations by Connett and Hirzy at last February’s Sydney anti-fluoride conference as sources here). As I pointed out in Connett fiddles the data on fluoride, this data actually does not show a strong relationship between IQ and fluoride. The figure (from Xiang et al., 2003) shows the relationship between IQ and urinary fluoride and, in this case, the fluoride explains only about 3% of the variance in IQ.

Despite being statistically significant (p=0.003) this is certainly not evidence for a causative relationship. Clearly other, unconsidered, factors contribute to the variance and if these were considered the relationship with fluoride may be non-significant.

(Readers may notice the figure uses data for urinary not drinking water fluoride. Unfortunately, Xiang did not give a similar figure for fluoride concentration in drinking water. I have contacted him requesting the similar data for drinking water but so far have not had a meaningful response. Xiang did report drinking water fluoride is well correlated with urine fluoride so the above figure probably gives a good idea of the variability in drinking water fluoride as well).

Connett and Hirzy effectively ignore the high variability in the data and rely on a trick to get this  second graph. By splitting the concentration range into groups and taking the mean IQ for each group they make the situation look a lot more respectable. Who would guess from this trick that fluoride only explained about 3% of the IQ variance?

Connett illustrates his next step with this slide.

He then claims that IQ deficits occur at a fluoride concentration of 1.26 ppm – he appears to have simply subtracted the value of one standard deviation from the mean of the lowest concentration group associated with a significantly different mean IQ to that of Xiang’s “control” group – Xinhaui village. That is strange because surely the first figure indicates  that low IQ values occur even for children with very low urinary fluoride, and most probably drinking water fluoride.

Connett then uses a safety factor of 10 (“to account for the wide range of sensitivity expected for any toxic substance in a large population”). Of course, this produces a maximum “safe” concentration of 0.13 ppm – which rules out all fluoridated water – and most natural water sources!

Connett goes on to promise his offsider, Bill Hirzy, will elaborate on the method they issued. Hirzy’s presentation did mention fluoride intake from other sources besides water. He then presents his conclusion on what the “safe daily dose” is fluoride – but no explanation of why! All the preceding slides in his presentation where self-justifying descriptions of his qualifications, employment history and how great his organisation, FAN, is.

Conclusions

Connett and Hirzy are claiming IQ deficits are more important than dental fluorosis for setting of maximum fluoridation levels in drinking water. They are campaigning to get this accepted by legislators and the EPA.

Connett has been promising publication in a scientific journal for several years and recently implied that a paper is under review. If their publication efforts are successful a more critical assessment of their approach will be possible.

Available information indicates Connett and Hirzy have no original data but are relying on data from a study of children in an area of endemic fluorosis in China. They are refusing to accept published information from areas where community water fluoridation exists.

Their analysis also appears to rely on a tricky processing of the data to obscure the fact that fluoride probably only explains about 3% of the variance in IQ measured by the Chinese researchers! Legislators and policy makers would be foolish indeed to make changes to fluoridation standards on the basis of such data and poor analysis.

I could, of course, be wrong so eagerly await the Connett & Hirzy (2016?) paper.

Similar articles