This was conducted by researchers at Adelaide and Queensland universities, which were linked in 2003 in promoting water fluoridation (Singh, 2003). The researchers are dental researchers, not experts in neurodevelopmental research. One of them is the infamous John Spencer.

This study makes claims that are not supported by the evidence presented, and unsupported claims.  The study clearly stretches the evidence to support and prove a
point, to “maintain confidence in the risk benefit balance of major caries-preventive programs using fluoride.”  The study did not have the capability to detect harm.

This study has significant limitations, some acknowledged by the authors. The study divides the study sample
into three subgroups, the lowest fluoride intake had 68 individuals due to costs and time-consuming nature of WAIS-IV and still give the study a high degree of confidence. The authors acknowledge the relatively small sample.  However, other limitations build on the small number.

• Fluoride exposure was estimated by community water fluoridation and dental fluorosis, both are crude
  measurements of total fluoride exposure. Socioeconomics is a reasonably accepted confounder for IQ and other confounders such as nutrition, regular schooling of good quality, pollutants like lead, and incidence of infectious disease also affect IQ. 
• Gender can also be a factor in Table 1.  The “zero” CWF cohort had 41.6% male and 58.4%  female; the middle subgroup 54% male and 46% female, and the 100% subgroup
  47% male and 53% female.
• Difference in parental educational was also significant with national education for Bachlors degree at 69% of
  the population, the “zero” fluoridation cohort had 78% tertiary degree (usually considered Bachlors or above) and 100% fluoridated had 86% Bachlors degrees or above. (Not all reports agree), significant 8% difference in educational attainment.
• Urban and rural residence was again significant with 100% fluoridated over 14% more Urban residence.
• Income difference of 32% vs. 50% is a significant difference and SES is well known to be a significant confounder to oral and general health.
• Fluoride supplements and total fluoride exposure have been more precisely measured in other studies. For example, bottled water consumption was not reported and is estimated at about half the water consumed in Australia and the USA.  Any obsession that fluoridated water is critical is simply not based on consumption facts.
• “Diffusion” or “halo effect” is discounted, although shipping processed foods from the city to the rural areas should not be discounted. 
• Bottled water is not considered.  Average water consumption in 2012-13 was about 1 litre/person/day.  About half the water consumed was bottled.
  Customer preference for a healthier alternative is one reason, although microplastics are a concern.
• Fluoride from the mother when ingested is not blocked by the placenta and is given to the developing foetus.
  Bone fluoride concentration also increases fluoride for the foetus during the third trimester as the mother’s bones are resorbed to give the foetus calcium. 

These and unknowns limit the confidence of the studies claim of “consistent evidence that childhood exposure to fluoride does not have effects on cognitive
neurodevelopment.”

The statistical reliability of a study is determined by the smallest cohort. This study did not have sufficient statistical power to detect harm with its small control sample
of 68 children without fluoridation exposure nor the 83 with dental fluorosis.

In contrast to the NTP-reviewed studies, this study did not measure individual-level total fluoride exposure or its best proxy, urine fluoride concentration. This study also did not measure prenatal fluoride exposure, a life stage sensitive to
developmental neurotoxicity.

Dental fluorosis as a measure of fluoride intake.

Do’s exposure measure of dental fluorosis may not provide a sufficiently precise proxy of true F exposure, and of relevant F exposure.

People vary greatly in their susceptibility to dental fluorosis.  The study’s analyses of IQ and dental fluorosis did not account for factors affecting dental fluorosis risk, including total intake, exposure timing, genetic variation,
metabolism, body weight, and nutritional factors [Alvarez 2009, Bhagavatula
2017, Huang 2008].

Also, they only measured fluorosis on the top front two incisors.  But even these two teeth can develop fluorosis from exposures over a fairly wide age range, so they may not be a reliable proxy for exposure during a fairly narrow age window. If susceptibility to IQ loss is heightened during a narrow age window, then Fluorosis may not correspond to that age window and be a relatively poor proxy for F intake at the susceptible time period. 

Several F-IQ studies have used both Fluorosis and other measures of F exposure, and several of them have found that the association with IQ is weaker with Fluorosis than with things like water F or urine F. 

Dental fluorosis, on the face of it, seems like an appealing measure of F exposure that could be relevant to developmental neurotoxicity, including IQ.  It tends to reflect total F intake and it can reflect somewhat narrow age ranges, depending on how the Fluorosis is measured and on which teeth.

But in reality, there seems to be a fair amount of variability in Fluorosis for any given F intake and timing of F intake.

This variability is equivalent to exposure measurement error, and is probably non-differential in most situations.  Non-differential exposure measurement error will always tend to produce bias toward the null.
So, any study that find’s no statistically significant effect (as Do’s) can potentially be explained as having suffered from too much exposure measurement error.  It may therefore be uninformative as to whether there is a real effect or not.

 The evidence brief concludes:

The use of the MIREC dataset for multiple analyses in multiple subgroups raises concerns
regarding the reliability of the findings and the relevance of the results to other populations. Repeated analysis of a single dataset will result in the identification of associations by chance. It is also possible that within the MIREC dataset, fluoride is a marker for a range of known causes of neurotoxicity, such as alcohol, pregnancy complications and sociodemographic variables. The over-reliance on a single dataset will provide a distorted view of the strength of possible associations.

There is no evidence to support any of the criticism. MIREC controlled for alcohol, for example. How can a population of Canadians not be relevant to comparable Western countries like NZ? Moreover, many of the reviews covered found evidence of harm. Yet the Evidence Review says it
supports earlier NZ reviews that fluoridation is safe.

By comparison, The U.S. National Toxicology Report ranked studies of fluoride’s developmental neurotoxicity and reported 18 of 19 high quality human studies up to May 1,
2020 reported lower IQ. The addendum further reported 12 of 12 high quality human studies between 2020 and 2023 also reported harm.

Of the three reviews cited by MoH as having no association with fluoridated water, one did find an association and one was fraudulent. 

Therefore only four of the five reviews are relevant and out of those, three did find an association with levels comparable to fluoridation and the other one found an association at higher levels. 

Gopu et al did find evidence of an association between fluoride in drinking water comparable to that used in NZ (see below). 
Kumar et al has been internationally recognised as fraudulent.
Miranda et al only found association at levels higher than fluoridation..

Taher et al found association between fluoride and reduced IQ.

Veneri et al found a consistent indication of a negative association between fluoride exposure and children’s intelligence, occurring from low fluoride concentrations when exposure was assessed through a biomarker (urinary fluoride), while some evidence of a threshold around 1 mg/L emerged from the pooled analysis based on drinking water fluoride.