Scaremongering graphic currently being promoted by Declan Waugh who is well known for misrepresenting the fluoride science
This graphic is typical of current anti-fluoride propaganda. It is scare-mongering, in that it is aimed at undermining community water fluoridation (CWF) which is accepted by health and scientific authorities as safe and effective. It relies on citations of recent research to give an impression of scientific credibility – but it misrepresents that research.
In fact, this research has produced confusing and contradictory results based on very weak relationships. Instead of the cherry-picking commonly indulged in by anti-vaccination and anti-fluoridation groups like this, all the findings in these studies must be considered. In this article, I have attempted to graphically present all these findings in one place. This makes clear how weak the evidence these activist groups rely on is and why it does not provide a basis for reviewing the current acceptance of CWF.
I have given below all the reported findings from the recent research (including the citations given by Declan Waugh in the above graphic). There is a lot here (I have not cherry-picked as the anti-fluoride activists do) so I present the findings graphically to provide a complete overview without the boring job of trying to understand detailed text. My apologies for the length of this article.
NOTE: I recommend readers refer to the cited papers for more details on methodology and definitions of the cognitive measures and F-exposure measures.
Does fluoridation influence IQ?
The answer provided by these modern studies is clearly no. Remember, these studies use data from areas where CWF is used or where drinking water concentrations are similar. They are quite different from the studies (previously relied on by anti-fluoridation activists) from areas of endemic fluorosis where fluoride intake is much higher and where many health problems result.
All the comparisons from fluoridated and unfluoridated areas in these studies are presented in Figure 1 below. The bars represent the standard deviations for the data sets and the data points are the means. A * indicates differences are statistically significant.
Figure 1: Comparison of IQ results in fluoridated and non0-fluoridated areas
The only statistically significant differences are for verbal IQ (VIQ) of 3-4 year-olds breastfed as babies (where the VIQ of children in fluoridated areas is higher) and for performance IQ (PIQ) of the same group (where the PIQ of children in fluoridated areas is lower). Till et al (2020) had to dig deep, use multiple measures of the cognitive score and subdivide the children into groups, to find an occasional difference. And these differences are contradictory.
I discussed the Till et al (2020) study which reported these occasional differences in Anti-fluoride propagandists appear not to read the articles they promote.
What about the relationships between IQ and measures of F intake?
Anti-fluoride propagandists ignore the data presented in Figure 1 above (and reported in the papers involved) and instead rely on cherry-picked relationships between measures of cognitive ability and various measures of fluoride exposure. Yes, some of these relationships, but only a small proportion, are statistically significant. But, importantly, none of these explain more than a few percents of the variation of the cognitive measure used.
Figure 2 below displays all the results from all the recent studies where linear regressions were used. The coefficient represents the size of the effect (eg., the change in IQ for every 1 mg/L increase of F exposure measure like drinking water F or maternal urinary F) and the bars represent the 95% confidence levels. The statistically significant (p<0.05) relationships are represented by red points while the green data points represent nonsignificant results.
Figure 2: Relationships of cognitive measures with exposure to fluoride obtained by linear regression analyses
Footnote: UF – concurrent urinary fluoride of the child. UFsg – UF adjusted using the specific gravity of urine. MUF – maternal prenatal urinary fluoride. MUFcr MUF – adjusted using urinary creatinine concentration. MUFsg – MUF adjusted using urine specific gravity. FSIQ – Full-Scale IQ. VIQ – Verbal IQ. PIQ – performance IQ. MDI – Mental development index.
Figure 3 below displays the results obtained by Barberio et al (2017) using logical regression of learning disabilities in children aged 3-12 years on urinary fluoride (UF), specific gravity adjusted urinary fluoride (UFsg), and creatinine adjusted urinary fluoride (UFcr). The data used was from two cycles of the Canadian Health Measures Survey (CHMS).
Findings for logical regression of ADHD and ADD on urinary fluoride are also included.
Figure 3: Relationships of cognitive measures with exposure to fluoride obtained by logical regression analyses
There are a lot of reported relationships in these two figures but only a few are statistically significant. Even these are contradictory – Thomas (2014) and Santa-Marina (2019) found positive coefficients while Bashash et al (2017), Thomas (2018), Green et al (2019) and Till et al (2020) reported some negative relationships. Barberio et al (2017) found a positive relationship for the data from combined CHMS cycles but this disappeared when UFsg or UFcr was used. Most of the reported relationships are not statistically significant.
Moving from nonsignificant to significant by adjusting urinary-F figures
This is illustrated by the evolution of the way the results are presented for the Thomas (2014) study which is related to the Bashash et al (2017) study. In this thesis and early conference reports (Thomas et al 2013 & Thomas et al 2014), She did not find any statistically significant relationships of child IQ with maternal urinary F (MUF) or maternal blood plasma F. But she did report a statistically significant relationship with MUFcr in her last conference paper (Thomas et al 2018).
So what happened?
There appears to be a change in the actual mother-child pairs used as indicated by the numbers and this sort of data selection can easily push a nonsignificant relationship into significance – especially when the relationship is so weak (see A conference paper on the maternal prenatal urinary fluoride/child IQ study has problems).
The other factor is that in the 2018 conference paper she has adjusted the MUF figures using creatinine concentration. Use of individual urinary fluoride measures, especially spot samples rather than a 24-hr collection, is a problem and is not a good measure of F exposure. Adjustment of urinary F using specific gravity or creatinine concentration is often used to improve the measure but this is problematic because creatinine concentration is influenced by a range of other factors. The adjusted MUF figures may actually be acting as a proxy for one of these other factors. This is why Barr et al (2005) recommended that:
“ For multiple regression analysis of population groups, we recommend that the analyte concentration (unadjusted for creatinine) should be included in the analysis with urinary creatinine added as a separate independent variable. This approach allows the urinary analyte concentration to be appropriately adjusted for urinary creatinine and the statistical significance of other variables in the model to be independent of effects of creatinine concentration.”
This is not done by any of the authors of these recent papers where urinary fluoride was used.
Thomas (2014) also reported a positive relationship of concurrent child urinary F (UF) with a cognitive measure, but not for girls. This seems to have been ignored in later reports – and by Bashash et al (2017) which used the same data but only reported the non-significant result for all children.
Till et al (2020) found that only relationships with PIQ were statistically significant. It is not clear why this happened considering no significant relationships were found for FSIQ or VIQ. It’s interesting that Till et al (2019) initially did not report the PIQ results and instead relied on a significant relationship of FSIQ with water F in children formula-fed as babies. Maybe the PIQ measurement is considered unreliable in practice. This finding was also heavily promoted by ant-fluoride campaigners – despite the fact that adjustment for other factors made this relationship nonsignificant (see Anti-fluoride propagandists appear not to read the articles they promote).
Most anti-fluoride campaigners have stuck with the initial FSIQ relationship – although a few who may have read the paper are now cherry-picking the PIQ relationships and ignoring the others.
What about fluoride and ADHD?
Three of these recent studies used linear regression when considering ADHD – but those of Malin & Till (2015) (claimed to be the first study to suggest an effect of fluoridation on ADHD) and Perrott (2018) are important. Not because one of the studies is mine – but because they illustrate a basic problem with correlation studies.
Even when multiple regression is used to adjust for covariants or other possible risk-modifying factors the investigation may miss an important risk-modifying factor. Not only does correlation not prove causation – the “significant” relationships themselves may be false when important risk-modifying factors are not included in the multiple regressions.
This happened with the Malin & Till (2015) study which reported statistically significant relationships of the extent of fluoridation in US states with ADHD prevalence. However, when the mean state elevation was included in the multiple regression of exactly the same data by Perrott (2018) the relationship with fluoridation extent disappeared (this had a p-value of 0.269 whereas those for Malin & Till 2015 were <o.o5). See Figure 4 below.
Figure 4: The effect of including other important risk-modifying factors on reported significant relationships
Figure 5 below shows the data reported by Bashash et al (2018) for the linear regression of a range of ADHD symptoms against urinary fluoride (UFcr).
Figure 5: Relationships of ADHD symptoms with exposure to fluoride obtained by linear regression analyses
The relationships were statistically significant for only four of the ten symptoms considered. Those relationships were very weak, explaining only a few per cent of the variance in ADHD prevalence (see Fluoridation and ADHD: A new round of statistical straw clutching).
The logistical regression results reported by Riddell et al (2019) for ADHD diagnosis and SDQ subscale score Urinary fluoride (UFcr) are given in Figure 6 below.
Figure 6: Relationships of ADHD diagnoses with exposure to fluoride obtained by logistical regression analyses
I discussed Riddell et al (2019) in my article ADHD and fluoride – wishful thinking supported by statistical manipulation?
This is another case where authors found unpromising results (no significant relationship for UFsg for example) and searched for other measures. It is also interesting that the significant relationships for water F and CWF status disappeared for younger children when age separation was used. The large confidence intervals in most cases indicate a large scatter in the data and very weak relationships.
I should also mention here the nonsignificant relationships reported by Barberio et al (2017) for ADHD and ADD (see Figure 3 above). These just underline how significant relationships are not common in these recent studies when looked at overall.
Update: Fluoride and sleep disturbances
Strictly, sleep disturbances don’t come under the classification of cognitive effects but a recent paper on fluoride and sleep disturbances is being promoted by anti-fluoride campaigners and should, therefore, be included here. For the sake of completeness.
I discussed the paper, Malin et al (2019), in my article Sleep disorders and fluoride: dredging data to confirm a bias. All the findings reported in that paper, and the supplementary files, are presented in Figure 7 below.
The authors report relationships of a range of sleep disorders against two measures of fluoride exposure – blood plasma-F and tap water-F. None of the relationships with blood plasma were significant (most had a p-value of 1.0). I discussed these in Sleep disorders and fluoride: dredging data to confirm a bias. and made the point that that bedtime and waketime were likely to be related to residence and the tap water F was simply acting as a proxy for regional location.
But again we see the use of a large number of measures for a “disorder’ and very few statistically significant relationships which are probably better explained by other factors than fluoride.
Considering all the findings together of the recent studies relevant to community water fluoridation and cognitive factors shows the results are weak, conflicting, and contradictory. This is probably not surprising considering the nature of the data (the studies were basically exploratory using existing databases – not designed experiments). Although adjustments were made for other possibly important factors this does not mean those really important ones (like the relationship of ADHD prevalence to elevation) were included. All the statistically significant relationships found were very weak – explaining a small proportion of the variance in the cognitive measure.
This is the sort of picture one might expect from exploratory studies using a large number of cognitive factors and measure of fluoride exposure. While these results may be useful in suggesting possible hypotheses to check in future better-designed experiments they are not sufficiently coherent to inform social health policy.
These recent studies do not provide sufficient evidence for revision of community water fluoridation policies because of possible effects on cognitive abilities. Anti-fluoride activists have only been able to use these studies in their scaremongering propaganda by cherry-picking results and ignoring the weakness of the relationships they cite.
Aggeborn, L., & Öhman, M. (2016). The Effects of Fluoride In The Drinking Water.
Barberio, A. M., Quiñonez, C., Hosein, F. S., & McLaren, L. (2017). Fluoride exposure and reported learning disability diagnosis among Canadian children: Implications for community water fluoridation. Can J Public Health, 108(3), 229.
Barr, D. B., Wilder, L. C., Caudill, S. P., Gonzalez, A. J., Needham, L. L., & Pirkle, J. L. (2005). Urinary creatinine concentrations in the U.S. population: Implications for urinary biologic monitoring measurements. Environmental Health Perspectives, 113(2), 192–200.
Bashash, M., Thomas, D., Hu, H., Martinez-mier, E. A., Sanchez, B. N., Basu, N., … Hernández-avila, M. (2017). Prenatal Fluoride Exposure and Cognitive Outcomes in Children at 4 and 6 – 12 Years of Age in Mexico. Enviromental Health Perspectives, 1, 1–12.
Bashash, M., Marchand, M., Hu, H., Till, C., Martinez-Mier, E. A., Sanchez, B. N., … Téllez-Rojo, M. M. (2018). Prenatal fluoride exposure and attention deficit hyperactivity disorder (ADHD) symptoms in children at 6–12 years of age in Mexico City. Environment International, 121(August), 658–666.
Broadbent, J. M., Thomson, W. M., Ramrakha, S., Moffitt, T. E., Zeng, J., Page, L. A. F., & Poulton, R. (2015). Community water fluoridation and intelligence: Prospective study in New Zealand. American Journal of Public Health, 105(1), 72–76.
Green, R., Lanphear, B., Hornung, R., Flora, D., Martinez-Mier, E. A., Neufeld, R., … Till, C. (2019). Association Between Maternal Fluoride Exposure During Pregnancy and IQ Scores in Offspring in Canada. JAMA Pediatrics, 1–9.
Malin, A. J., & Till, C. (2015). Exposure to fluoridated water and attention deficit hyperactivity disorder prevalence among children and adolescents in the United States: an ecological association. Environmental Health, 14(1), 17.
Malin, A. J., Bose, S., Busgang, S. A., Gennings, C., Thorpy, M., Wright, R. O., … Arora, M. (2019). Fluoride exposure and sleep patterns among older adolescents in the United States : a cross-sectional study of NHANES 2015 – 2016. Environmental Health, 1–9. Retrieved from https://link.springer.com/article/10.1186/s12940-019-0546-7
Perrott, K. W. (2018). Fluoridation and attention deficit hyperactivity disorder a critique of Malin and Till (2015). British Dental Journal, 223(11), 819–822.
Riddell, J. K., Malin, A., Flora, D., McCague, H., & Till, C. (2019). Association of water fluoride and urinary fluoride concentrations with Attention Deficit Hyperactivity Disorder in Canadian Youth. Submitted to Environment International, 133(May), 105190.
Santa-Marina, L., Jimenez-Zabala, A., Molinuevo, A., Lopez-Espinosa, M., Villanueva, C., Riano, I., … Ibarluzea, J. (2019). Fluorinated water consumption in pregnancy and neuropsychological development of children at 14 months and 4 years of age. Environmental Epidemiology, 3.
Thomas, D. B. (2014). Fluoride exposure during pregnancy and its effects on childhood neurobehavior: a study among mother-child pairs from Mexico City, Mexico. University of Michigan.
Thomas, D., Hu, H., Basu, N., Sanchez, B., Bellinger, D., Schnaas, L., … Tellez-Rojo, M. M. (2013). A prospective study of prenatal exposure to fluoride and neurobehavior : preliminary analyses. Environmental Health Perspectives.
Thomas, D., Hu, H., Basu, N., Martinez-Mier, E. A., Sanchez, B., Bellinger, D., … Tellez-Rojo, M. M. (2014). Urinary Fluoride in Pregnant Women and Prenatal Fluoride Exposure and Mental Development Index ( MDI ) in 1-3 Year Old Infants from Mexico City, Mexico. Environmental Health Perspectives, 1(Icc), 2–3.
Thomas, D., Sanchez, B., Peterson, K., Basu, N., Angeles Martinez-Mier, E., Mercado-Garcia, A., … Tellez-Rojo, M. M. (2018). OP V – 2 Prenatal fluoride exposure and neurobehavior among children 1–3 years of age in Mexico. Environmental Contaminants and Children’s Health, 75(Suppl 1), A10.1-A10.
Till, C., Green, R., Flora, R., Hornung, R., Martinez-Mier, E., Blazer, BFarmus, L., … Lanphear, B. (2019). Fluoride Exposure from Infant Formula and Child IQ in a Canadian Birth Cohort. Environmental Epidemiology, 3.
Till, C., Green, R., Flora, D., Hornung, R., Martinez-mier, E. A., Blazer, M., … Lanphear, B. (2020). Fluoride exposure from infant formula and child IQ in a Canadian birth cohort. Environment International, 134 (September 2019), 105315.