A useful reminder of how activists lie with statistics
The local anti-fluoride people have not stopped rabbiting away with their “science” – searching for anything bad they can argue is caused by community water fluoridation. The latest claim they make – fluoridation is responsible for infant deaths!
That table is simply lifted from a conspiracy style web page Why Pregnant Women Should Not Drink Fluoridated Water.
OK – these statistics might fool some people – especially if you have a bias to confirm. But the more critical person might pick up that these figures are most likely cherry-picked and want to see the full data set or some analysis of the data.
It’s not hard to find this data as there are tables of all sorts of things for US people organised by state. So, is there a relationship between infant deaths in each state and the extent of fluoridation in each state?
There actually does appears to be one at first sight – here is the graph of the data for infant deaths in 213 plotted against the extent of fluoridation in 2012.
But, just a minute – it is not actually statistically significant (p=0.106) and would account for only about 5% of the variance in infant deaths. Fluoridation is certainly not the main factor – and probably involved at all if other factors are considered.
Here I will just take into account the influence of state elevation – because I know from previous work that fluoridation extent is related to mean state elevation (see ADHD linked to elevation, not fluoridation).
Here is that relationship for the extent of fluoridation in 2012:
So, the extent of fluoridation in each state is related to mean state elevation and this relationship is statistically significant (p=0.005). Actually not surprising as the larger and older cities where fluoridation might be expected are generally situated at lower elevations for a number of reasons.
But what is the relationship between infant death and mean state elevation? Well, it is stronger than for the extent of fluoridation (p=0.002). Elevation accounts for about 18% of the variance in infant deaths in 2013.
Finally, let’s combine both elevation and extent of fluoridation into a multiple regression and see what the relationship when both factors are combined.
This multiple regression shows a statistically significant association (p=0.007) of the extent of infant deaths in each state in 2013 with the mean state elevation. However, there is no statistically significant association (p=0.592) with the extent of state fluoridation.
So while infant deaths could be explained by mean state elevation and most probably one or more other factors, they certainly are not explained by the extent of fluoridation. Not at all!
Preterm birth and conspiracy theory
In her article, Karen Favazza Spencer makes the bald claim “Fluoridation is positively correlated with preterm birth and increased death rates by state “ – again citing from the conspiracy style web page mentioned above.
Sure, that page makes that claim – “Domestic water fluoridation was independently associated with an increased risk of PTB [preterm birth].” But that is hardly credible evidence because that page goes on the say:
“This study was never published nor was any follow-up research done, despite the fact that 2 years earlier, the US Institute of Medicine reported: ‘Those born preterm have an appreciable risk of long-term neurological impairment and developmental delay.'”
Strange! It is not hard to find data for preterm birth. In fact, here it is for 2014 compared with the extent of state fluoridated in 2012:
Clearly, there is no association between preterm births and extent of fluoridation. Yet that web page claimed there was and that the information had been suppressed!
I guess that is another way ideologically motivated activists “prove” these sort of things – invoke a conspiracy theory to claim a relationship exists but the data is suppressed.
So, once again the lesson is – never take at face value the claims made by anti-fluoridation activists – no matter how “sciency” their information looks or what data they invoke to “prove” them.
Always check such claims for yourself.