Home > skepticism, Uncategorized > fluoride fallacies: dead-end consensus

fluoride fallacies: dead-end consensus

Before i delve further into the misconferral of authority by antifluoridationists, an i’d like to make point about scientific consensus that i neglected in my previous post. It is easy to think of scientific knowledge as essentially static, with the occasional fact or theory being overturned when new evidence surfaces. We imagine scientific knowledge as a sort of pyramid, with the most reliable knowledge composing the base and provisionality increasing as we move upward. While this model is useful for making the quintessentially (Humean–)Bayesian–Pricean point that stronger evidence should more strongly inform our beliefs, it falls apart as soon as we widen our scope from the single Providence province to which the pyramid aspires. The sciences are not independent avenues of discovery, as imagery of a field of pyramids might suggest, but highly interdependent configurations of highly intradependent evidences and interpretations.

To draw an analogy of my own: The highly symbolic and culturally entangled concept of gender is not reducible to the space determined by orthogonal (and necessarily binary) spectra of identity, expression, “biological” sex, and attraction; it involves the highly nontrivial and individualized interplay of these factors, each of which in turn arises from the interplay of several distinguishable (if not wholly distinct) factors. A perhaps preferable model of gender is the graph representation of these interrelations: a node for each factor and edges tying them together, with the understanding that any particular node (say, “attraction”) may dissolve into a subnetwork of factors (intimacy, arousal, sexuality, satisfaction, saturation, etc.) on closer scrutiny.

Similarly, out of the intricate network of implications, corroborations, constraints, and tensions that connect elements of our aggregate body of facts and interpretations arises a web of knowledge. Moreover, this web may be anywhere localized, as though sliding a magnifying lens over a paper map, so that all our knowledge may be interpreted in terms of its relevance, or “consequential proximity”, to one’s topic of choice. Ultimately no particular discipline or theory is more “central” than any other.

The upshot of this view for an understanding of scientific consensus and “established facts” is that “facts” are never finally established at all; they remain perpetually provisional. Yes, they are “falsifiable”, but the simplistic notion that a single experiment or discovery could overturn a massively confirmed theory is naïve. The famous 1919 light-bending experiment alone no more overturned Newtonian physics than the 2011 neutrino-firing experiment alone overturned special relativity. The former was informed by a mature theory with rigorous mathematical foundations that explained other known anomalies and has been confirmed by subsequent experiments since; the latter was isolated and unprecedented, and an explanation consistent with existing knowledge was uncovered within a year.

The corollary upshot is that knowledge that survives does so amidst an unrelenting storm of “attempts” at disconfirming it. After a while these cease to be targeted attempts at disconfirmation; but to the extent that ongoing research relies upon existing knowledge to make sense of its results, this existing knowledge becomes part of the overall framework being subjected to experimental attack.

Now, it is a serious concern even in many sciences that our biases and expectations prevent us from exploring potentially vital avenues of inquiry or noticing problems with reigning methodologies. As commenter bruce at Dan’s blog puts it,

a strong consensus influences what research is done and what evidence is discovered. When all the department chairs, journal editors, and government grant givers are supporters of an idea, a researcher is wise not to dissent.

Antifluoridationists play up this fear, suggesting that “established” results, like the hardening and remineralizing effects of waterborne fluoride on enamel, originally demonstrated using measurements and statistics contemporary to the 1940s, are not being subjected to the improved scrutiny of modern scientific methods, and by implication would not survive it. Contrary to their insistence, however, this dead-end view of consensus is fundamentally misguided. The preventative effect of water fluoridation on dental caries is central to the design of ongoing research, and thereby is reaffirmed whenever this research returns meaningful results.

For example: It has been well-established over the decades that children living in communities that adjust the fluoride concentration in their water to optimal tend to suffer less tooth decay, though the effect is subject to other influences and difficult to quantify. However, the benefits of community water fluoridation (CWF) for adults were until recently a matter of some contention. According to the authors of a recent study of Australian adults,

Before 1990, few studies investigated the caries-preventive effects of fluoridation in adults (Rugg-Gunn and Do, 2012). This probably reflected a view, now outdated (Beltran and Burt, 1988), that fluoridated water was effective only when ingested prior to tooth eruption.

In particular because the distribution of benefits plays an important role in policy discussions, this effect deserved to be better understood.

The authors conducted a study specifically to tease out this effect. This required a huge (and representative) sample of Australians (ultimately 3,779 participants) to get their teeth examined and fill out a residential history; a team of dentist-examiners to count their decayed, missing, and filled teeth or tooth surfaces; knowledge of the fluoride content of various locations’ water supplies over time; and a statistical analysis that used this information to assign each participant a “fluoridation exposure” history, taking several possibly confounding factors, like age, income, and having actually gone to the dentist regularly, into account. Participants’ fluoridation exposure histories were grouped into four lifetime percentage ranges: 0–25% (“negligible”), 25–50%, 50–75%, and 75–100% (“prolonged”). The central question was, on average, and once all known factors have been accounted for, how many more teeth or tooth surfaces have suffered decay among people having negligible histories versus people having prolonged histories?

The authors found a substantial difference by both measures, amounting to 10–11% fewer decayed, missing, and filled teeth and 21–30% fewer decayed and filled tooth surfaces in the prolonged exposure group relative to the negligible exposure group. Moreover, measurements taken from the partial-exposure groups tended to lie between those taken from the diametric groups, suggesting a dose-response relationship and lending support to a causal explanation.

These ranges in caries reduction aren’t due to uncertainty. The authors divided their participants into two cohorts to see whether their benefits differed. The 21% figure was measured in the cohort of participants who were born after 1960, about the time that CWF was introduced to Australian communities on a large scale, while the 30% reduction was measured in the cohort born before 1960. These cohorts were measured separately, the authors note, “with the expectation that the association would be stronger in the 1960–1990 cohort”. This is because the established preventative effect of CWF during childhood would only have impacted the younger cohort. Indeed, caries rates for the younger cohort were about half that of the older cohort, leaving less room for a percentage reduction due to lifetime exposure. The study therefore quantified both the combined effect of childhood following the large-scale implementation of CWF with lifetime adult exposure (in the younger cohort), and that of adult exposure without widespread CWF during childhood (in the older cohort).

So, while the study’s main contribution is to shift the balance of evidence regarding the impact of fluoridation on adults, proper methodology demanded that it take our existing knowledge of the benefits of CWF to children into account.

Another Australian study, also published this year, concerned the effect of sugar-sweetened drinks like Dr. Pepper soda on caries in children, itself surprisingly an underinvestigated association. To account for the confounding effect of CWF, the authors similarly included for each child participant (16,857 total) a percentage lifetime exposure to optimally fluoridated water, as well as stratifying and correcting for a similar range of confounding factors as the previous study.

The authors measured an average additional .34 decayed, missing, or filled deciduous (baby) teeth among children who drank 1 or 2 sugary drinks per day versus those who drank none, and an additional .46 among children who drank 3 or more. The trend was weaker but still easily detectable in permanent teeth. Within each bracket, children with more lifetime exposure to CWF fared better than children with less. All this was to be expected. Less expected was that among children with over 50% lifetime exposure to CWF, the increase in caries due to sugary drinks disappeared—it became too small to detect. That is quite a preventative effect. Also worth noting is that the contribution of CWF far outweighed that of (reported) toothbrushing and even of household income. Both observations will better inform policymakers—and, one hopes, voters—when they decide whether to introduce or preserve optimal fluoridation in their communities; and both observations were made possible by an intelligent study design that took existing knowledge into account.

So it goes in science: As new niches are explored, methods rely increasingly on prior results; every time these methods are successful, the prior results are again vindicated.


It is worth pointing out the more sophisticated objections raised by Dan Steinberg of Fluoride Free NRV. In his original flier, he writes that

The observed [cavity protection of ingested fluoride—presumably meaning CWF generally—is] likely not even real because fluoride delays emergence of the permanent teeth. Delayed tooth eruption simply postpones cavities, creating a false observation of cavity prevention. This effect has NOT been accounted for in fluoridation studies.

Neither recent study discussed above was designed with this phenomenon in mind (presumably because it is not recognized as a serious problem in the scientific community), but both defy it as an alternative explanation: The study of adults was specifically designed to measure the effect of CWF during adulthood, after permanent teeth have erupted. Meanwhile, the study of children and sweetened drinks included measurements both on deciduous and on permanent teeth, and in both cases found a sizable reduction in caries among children in fluoridated locations. We’ll see if the local campaign changes its rhetoric on this front, but the larger anti-fluoridation movement will most assuredly not.

So it goes in pseudoscience: As new research undermines favored hypothesis, either the hypotheses are finagled into conformance or the new research is ignored.

Interestingly again from a policymaking perspective, the authors noted “that children with higher [sugar-sweetened drink] consumption brushed their teeth less frequently, which is an example of how risk behaviors tend to cluster together in many at-risk individuals”. This is a commonly neglected point in discussions over water fluoridation, which is poised to confer greater benefits on those who need it most.

  1. No comments yet.
  1. No trackbacks yet.

Enlighten me.

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: