Another study comparing effect in children from nonendemic areas (Dagang) and endemic fluorosis area (Jinghai) about 80km apart in the Tianjin area of China. Hardly sampling the same population.
Anti-fluoride campaigners still use studies from endemic fluorosis area of China in their campaign against community water fluoridation (CWF). This misrepresents the situation. While the control areas they use (non-endemic fluorosis areas where there are no health effects) are relevant to CWF the data from the endemic fluorosis areas where fluoride intake is high are simply not relevant to CWF.
However, some of these studies provide enough information to enable checking for health effects in the control, non-endemic fluorosis areas. I discuss one of these studies below – Cui et al (2018) – and show it finds no health effects at the low fluoride concentration relevant to CWF. The use of this study in anti-fluoride campaigns is therefore misleading.
I also show how even in his misrepresentation of this study the “Research Director” of the Fluoride Action Network (FAN), Chris Neurath, makes basic mathematical mistakes.
The citation for this study is:
Cui, Y., Zhang, B., Ma, J., Wang, Y., Zhao, L., Hou, C., … Liu, H. (2018). Dopamine receptor D2 gene polymorphism, urine fluoride, and intelligence impairment of children in China: A school-based cross-sectional study. Ecotoxicology and Environmental Safety, 165(August), 270–277.
The study used school children (7 to 12 years old) from the districts of Jinghai and Dagang in Tianjin of China (see map above). The endemic fluorosis area had drinking water fluoride concentrations of 1.52–2.49 mg/L and the nonendemic fluorosis area had drinking water fluoride concentrations of 0.20–1.00 mg/L.
Of course, the results for children in the endemic fluorosis area are simply not relevant to CWF where the drinking water concentrations are usually less than 0.8 mg/L.
ONLY THE DATA FROM THE NON-ENDEMIC FLUOROSIS AREA IS RELEVANT TO CWF.
Child IQ values were compared with urinary fluoride levels and regression analysis used to determine if there was any relationship. The children were also tested for the presence of different genes and each of three different genetic groups was considered separately.
The data from both endemic and non-endemic fluorosis areas were combined. That for urinary F was skewed and had to be transformed to provide the normal data distribution required for linear regression analysis. Consequently, the authors report linear regression results for the relationship of child IQ to the logarithm of urinary F [Log(UF)].
No statistically significant relationships were found for two of the genotypes (designated CC & CT) but there was a significant relationship for the genotype designated TT. Less than 14% of the children had this gene.
Figure from Cui et al (2018) showing a negative relationship of child IQ with the logarithm of urinary fluoride. Equation of fitted line is IQ = 117.48 – 9.75*Log(UF). R-squared = 0.142, p = 0.012. 95% CI for coefficient -17.21,-2.29
So, of course, the anti-fluoride campaigners simply go with the results for the children with the TT gene and ignore the results for the other 86% of children.
Nothing new here – they always ignore results that they can not use in their campaigns to confirm their biased presentations.
Analysis of the data for the low fluoride levels relevant to CWF doesn’t support activist claims
Here I will just consider the data relevant to CWF. After digitally extracting the data from the figures in the paper I restricted linear regression analysis to the children with urinary F values of less than 2 mg/L (which is still rather high for areas where CWF is used). The figure below displays that data, together with the results of linear regression analysis. Untransformed values were used because data for urine fluoride concentrations less than 2 mg/ml is normally distributed.
Linear regression analysis for the relationship of child IQ with urinary F. Children with TT gene about 44% of the sample. Children with CC and CT genes about 86% of the sample.
There were no statistically significant (p < 0.05) relationships, either for the TT variant (red triangle) or for the other variants (open circle) (CC & CT combined in this figure).
So, once again we see that if the appropriate data from these studies are used they confirm that there is no relationship of child IQ with measures of fluoride exposure at concentrations relevant to CWF (see also New study touted by anti-fluoridation campaigners actually indicates fluoridation is safe).
FAN’s “Research Director” makes simplistic mistakes
FAN plans to use studies like Ciu et al (2018) in their upcoming case against the US Environmental Protection Agency. Their aim is to attempt to establish cognitive effects as the main harms from CWF and then use studies like these to argue against CWF. Even though studies like this simply establish that there is no harm from fluoride concentration used for CWF. One hopes that the experts testifying for the EPA show how these studies are misrepresented by FAN.
FAN has provided a presentation by their “Research Director,” Chris Nerath, which they claim summarises their arguments. It’s a “pretty” PowerPoint presentation (FAN describes it as “powerful“) and may fool some people, but it just does not stand up to scientific scrutiny.
Quite apart for the misrepresentation of these scientific studies, and use of studies like Cui et al (2018) which are not relevant to CWF, Neurath simply makes basic scientific mistakes.
For example, in his slide 33, he claims that Cui et al (2018) showed a 10 point IQ loss for a 1 mg/L increase in urinary F. But that is simply not true – his mistake is that he ignored the fact that the linear relationship reported by Cui et al 2018) [IQ = 117.48 – 9.75*Log(UF)] is based on log values of urinary F where the value of the coefficient is -9.75. In other words, the 10 point loss is for an increase of urinary by a log value of 1 is equivalent to an increase from a concentration of 1 mg/L [Log(1) = 0] to a urinary F concentration of 10 mg/L ([log(10)=1].
Slide 33 from the presentation by Chris Neurath, FAN’s “Research Director” displaying an embarrassing mathematical mistake.
Let’s do the correct calculation for him. The Cui et al (2018) relationship shows an IQ value 117.48 for a urinary F concentration of 1 mg/L [Log(1)=0] and 114.55 for a urinary F concentration of 2 mg/L [Log(2)=0.301]. So the loss is only 2.9 IQ points.
Neurath thinks it sounds much better for his case to say a loss of 10 IQ points but all he has done is shown either he did not read Cui et al (2018) properly or does not understand a simple mathematical relationship.
However, I should stress Neurath’s argument is irrelevant to CWF, as well as being mathematically wrong because its analysis includes data from endemic fluorosis areas. In fact, there is no statistically significant relationship between child IQ and urinary fluoride either for the overall group or for the separate genetic groups considered.
The graph in Neurath’s slide is adapted from Cui et al (2018) which made an attempt to determine a “safety threshold of urine fluoride levels for IQ
impairment” for the children with the TT gene. They defined this “safety threshold” as the urinary fluoride value corresponding to the mean IQ for the group. This seems arbitrary to me and the authors make no attempt to justify the definition.
They then divided the data into quintiles according to urinary fluoride values. Quintile 4 was the first quintile where the mean IQ value is below the mean IQ for the whole group so they assumed to mean log(UF) for that quintile to represent the “safety threshold.”
It all seems quite hairy to me – but I suppose the method produces a few data points which Neurath was able to plot on a graph and make to appear impressive. But look at the spread of the data (Neurath does not show this) in the graph below where the bars represent the spread of data for each quintile and the dotted line is the mean IQ value for the group. The spread is hardly surprising – the overall data is very scattered and only about 9 data points were used for each quintile.
I frankly think this method of determining a “safety threshold” is meaningless. We could do exactly the same with the data where urinary fluoride is less than 2 mg/L – much more relevant to CWF. This is the result – using quartiles of about 8 data points each.
There are many other misrepresentations and mistakes in Chris Neurath’s Powerpoint presentation and I may return to some of them later. However, let’s hope the court recognises these and rejects FAN’s attempts.
Anti-fluoride activists continually use studies from areas of endemic fluorosis in their campaigns against CWF. However, when the actual data relevant to community water in these studies are considered they usually show no health effect. There is no doubt that people living in endemic fluorosis areas suffer a range of health problems. But where these studies provide complete data they almost always could be used to support CWF.
Activists like FAN and their “Research Director” simply clutch on to any study they can finds which appears to show harmful effects of fluoride and ignore the fact that they are hardly ever relevant to the fluoride concentration used in CWF. They cherry-pick and ignore, or cover-up, any information not supporting their bias.
This approach is hardly scientific. It is not objective and never undertakes a critical review of the studies used. These activist display a thoughtless approach to scientific research when they opportunistically use scientific studies like this. Their approach is unthinking and it is hardly surprising that they make simple mathematical and statistical errors like the one described here and made by Paul Connett recently (see Anti-fluoridation propaganda now relies on only four studies. 4: Till et al (2020) and Author confirms anti-fluoridation activist misrepresentation of her work).