by Paul Connett
Typically, to take into account intraspecies variation, the lowest level or dose at which toxicity is observed (i. e. , lowest observable adverse effect level, or LOAEL) is divided by 10 to set the margin of safety. This factor of 10 assumes that the most sensitive person is ten times more sensitive than the least sensitive.
Sometimes more conservative regulatory agencies insist on working from a no observable adverse effect level, or NOAEL. If that is not available, they require that a margin of safety of 100, not 10, be applied to the LOAEL.
Sometimes a margin of safety lower than 10 is used if data have been collected on an adequate range of doses from studies in very large populations. In such a case it is assumed that enough people have been observed to cover the full range of human sensitivity. However, in the case of fluoride, many of the studies finding adverse effects have involved relatively small study groups. Such studies would not cover either the full range of anticipated sensitivity or the full range of dose when a much larger population is exposed.
The people who need special protection in the case of exposure to fluoride include the very young, the very old, those with poor diet (including borderline iodine deficiency), those with poor kidney function (which reduces the ability to excrete fluoride), those who consume above-average quantities of water (athletes, diabetics, etc. ), and infants who are fed formula that has been reconstituted with fluoridated water.
Based on current levels of exposure and the levels at which effects were shown to occur in the 2006 NRC report, 8 it is hard to see how a scientifically defensible safety factor could yield a safe level for fluoride in water of more than 0. 1 ppm. Indeed, Dr. Robert Carton, a former risk assessment specialist at the Environmental Protection Agency (EPA), has argued, based on the NRC findings, that the maximum contaminant level goal (MCLG) for fluoride should be set at zero, as has been done for both lead and arsenic. 9 This is what we believe a scientific margin-of-safety analysis would show.
Those who claim that the practice of water fluoridation is safe for everyone have a clear obligation to demonstrate that by performing a careful margin-of-safety analysis for the adverse health effects reviewed and summarized in the 2006 NRC report. 10 Such a demonstration would have to include the rationale for choosing the most appropriate (i. e. , most sensitive) end point and LOAEL for that end point (the end point being a known or reasonably anticipated health effect) for all those health effects discussed by the NRC report.
It is reckless to continue promoting fluoridation when studies indicate that thyroid function may be lowered at 2. 3 ppm, 11 IQ in children may be lowered at levels as low as 1. 9 ppm12, 13 or even at 0. 9 ppm if there is borderline iodine deficiency, 14 and hip fractures in the elderly may be increased at levels as low as 1. 5 ppm and tripled at levels over 4. 3 ppm. 15, 16 Unless all these effects can be dismissed because all the relevant studies have been shown to be fatally flawed, there is clearly no adequate margin of safety to protect the whole population from those end points. Thus, the practice of fluoridating the public water system should be discontinued. Furthermore, if the relevant studies have faults but still raise plausible doubts about safety, the program should be stopped until research yields reliable conclusions (the precautionary principle; see chapter 21).
Not a New Idea
The application of a margin-of-safety analysis to water fluoridation is not a new idea. As long ago as 1956, Benjamin Nesin, a prominent water engineer, indicated that the safety factors discussed at the time offered inadequate protection. Focusing particularly on the range of doses one could anticipate across the whole population, he wrote the following:
The proponents have tried to demonstrate various factors of safety which are patently naive. . . It has been customary to consider a minimal factor of safety of not less than 10 for substances which are admitted to water supplies. This would mean that ten times the amount of the proposed substance when present in the water supply would be definitely without harm to human or beast. It is obvious from the knowledge of fluoride toxicity that such factor of safety cannot be established when fluoride is added to the public water supply at the level recommended by the proponents of fluoridation. In view of the fact that the range of water consumption may vary over a ratio of 20 to 1 the insistence of a safety factor of 10 is exceedingly moderate. 17
Inadequate Margin-of-Safety Calculations
To date, the margins of safety used by agencies in the United States (and other fluoridating countries) for fluoride have been woefully inadequate. We examine a few of these.
EPA and the MCLG
The U. S. EPA’s Office of Drinking Water’s 1986 derivation of the 4 ppm maximum contaminant level goal (MCLG) for fluoride in drinking water18 is scientifically flawed. The EPA’s Office of Prevention, Pesticides and Toxic Substances explained the derivation of the current MCLG by the EPA’s Office of Drinking Water as it was preparing to permit sulfuryl fluoride as a new food fumigant:
For fluoride, both the MCL and the MCLG have been set at 4. 0 ppm in order to protect against crippling skeletal fluorosis. The MCLG was established in 1986 [Federal Register 51, no. 63] and is based on a LOAEL of 20 mg/day, a safety factor of 2. 5, and an adult drinking water intake of 2 L/day. The use of a safety factor of 2. 5 ensures public health criteria while still allowing sufficient concentration of fluoride in water to realize its beneficial effects in protecting against dental caries. The typical 100x factor used by the HED [the EPA’s Health Effects Division] to account for inter- and intra-species variability have been removed due to the large amounts of human epidemiological data surrounding fluoride and skeletal fluorosis. 19
Having explained this derivation of the MCLG of 4 ppm, the Office of Pesticides then used it as a basis to determine tolerances for fluoride residues left on foodstuffs treated with sulfuryl fluoride. We discuss the way they did this in three different health risk assessments in the section “EPA Pesticide Division” below. But first let us break down the 1986 derivation of the MCLG by the EPA’s Office of Drinking Water into its component parts:
1. The end point chosen was crippling skeletal fluorosis.
2. The LOAEL offered for this was 20 milligrams per day.
3. The safety factor offered to protect the most vulnerable was 2. 5.
4. The amount of water drunk per day was 2 liters.
5. The only source of fluoride considered was water.
The EPA’s (Office of Drinking Water) calculation was as follows: 20 milligrams per day divided by 2. 5 (safety factor) equals 8 milligrams per day. If one assumes that a person drinks 2 liters of water a day, the supposed safe level is 4 milligrams per liter, because if someone drank 2 liters of water at that level, they would receive the supposed safe level of 8 milligrams of fluoride. Thus the EPA arrived at an MCLG of 4 ppm.
There are six things wrong with the EPA’s derivation. It is hopelessly wrong at all five steps, and two mistakes are made on one of those steps. Even at the time, the inadequacies of the derivation were pointed out by some EPA scientists, including Dr. Robert Carton, who claimed that it was manipulated for political reasons. 20 Today, with the benefit of more research findings, the inadequacies are even more glaring. We now examine each step of the calculation and the six places where the EPA went wrong.
Mistake one: Selecting crippling skeletal fluorosis as the end point of concern.
There are two problems with the EPA’s selection of crippling skeletal fluorosis as the only end point of concern. First, even if the EPA wished to focus on fluoride’s impact on the bone it was both unscientific and contrary to common sense to focus solely on the crippling phase of skeletal fluorosis. More sensitive end points were known in 1986 and these include the earlier stages of skeletal fluorosis as well as bone defects (see chapter 17). Second, in 1986, there was evidence available that fluoride could have impacts on thyroid function at levels far lower than those that caused crippling skeletal fluorosis21, 22 (see chapter 16).
As far as skeletal fluorosis was concerned, a lot
of observable damage occurs to the bone and connective tissue before the crippling phase is reached. The existence of arthritic symptoms in the pre-crippling stages of the disease has been widely reported in the literature prior to the EPA’s 1986 determination of the MCLG23–28 (see chapter 17). While not everyone with pre-crippling clinical fluorosis will experience arthritic pain, the evidence is clear that some people will. Thus the EPA should have selected pre-crippling clinical effects as a more sensitive end point of concern. Indeed, the 2006 NRC panel recommended to the EPA that stage II of skeletal fluorosis be considered an adverse health effect (see chapter 14).
According to the NRC, “The committee judges that stage II is also an adverse health effect, as it is associated with chronic joint pain, arthritic symptoms, slight calcification of ligaments, and osteosclerosis of cancellous bones. ”29
In addition, as far back as 1956, researchers had found that cortical bone defects were significantly higher in children in fluoridated Newburgh, New York, compared to unfluoridated Kingston30 (see chapter 17).
Mistake two: Using a LOAEL of 20 mg/day
If the EPA had used cortical bone defects as the most sensitive end point then the LOAEL should have been selected between 1 and 2 mg/day. This assumes that children in fluoridated Newburgh were drinking one to two liters of water per day and that water was their main source of fluoride. 31
If the EPA had selected lowered thyroid function as the most sensitive end point, then the study by Bachinskii et al. would indicate a LOAEL in the range of 2. 3 to 4. 6 mg/day. 32 This again assumes that those who had experienced lowered thyroid function drinking water at 2. 3 ppm fluoride were drinking between one to two liters of water per day.
If the EPA had selected a pre-crippling stage of skeletal fluorosis it is hard to see how they could have selected a LOAEL larger than 10 mg per day and probably considerably less. So whichever adverse health effect the EPA had selected, of the several options available, it is clear that 20 mg per day was far too high to be protective.
Mistake three: Failure to use an adequate safety factor.
Even if we ignore the inadequate LOAEL used by the EPA, the fact remains that the agency failed to apply an adequate safety factor to even that LOAEL to allow for the range of vulnerability to any toxic substance in a human population.
To claim that the normal safety factor of 100 was dropped because of “the large amounts of human epidemiological data surrounding fluoride and skeletal fluorosis” is wrong on two fronts. First, the EPA ignored the large amount of data on this issue from India and China, which have areas where both dental and skeletal fluorosis are endemic. Bone effects have been observed in India at levels of fluoride in water ranging from 1 ppm to 3 ppm where there is poor nutrition. 33 Second, the agency derived the LOAEL of 20 mg/ day largely from Kaj Roholm’s work, which was based on a small sample of otherwise healthy male industrial workers. 34 One needs a much larger safety factor to cover the whole range of sensitivity anticipated in a large human population.
Mistake four: Dropping the normal safety margin to protect the fluoridation program.
The EPA pesticide division, in the quote above, made it very clear why the standard protective safety factor of 10 or 100 was sacrificed in the EPA water division’s calculation, stating that it was to allow “sufficient concentration of fluoride in water to realize its beneficial effects in protecting against dental caries. ”35 In other words, the EPA admitted that its derivation involved protecting the water fluoridation program. This kind of thinking may have been appropriate when the agency moved from an MCLG (a goal) to an MCL (a standard), but it was inappropriate to use such reasoning in determining the MCLG. For an MCLG, the task is to determine an ideal safe level based on scientific studies of toxic end points and to apply an appropriate safety factor to protect the most vulnerable. The EPA should not have allowed the purported benefit of the fluoridation program to interfere with what should have been a scientific determination of the ideal goal to protect everyone from “known or reasonably anticipated health effects, ” as mandated in the Safe Drinking Water Act.
It should be noted that, as requested by the EPA, 36 the 2006 NRC report37 ignored any discussion of the benefits of the water fluoridation program when the NRC panel members set out to provide the toxicological data that could be used by the EPA to determine a new MCLG.
There are two more problems when we move from the supposedly safe level of 8 milligrams per day (20 mg/day divided by 2. 5) to an MCLG of 4 ppm.
Mistake five: Assuming that people drink only 2 liters of water a day.
The assumption that people drink only 2 liters of water per day is clearly wrong. Although that may be the average consumption, the MCLG has to protect everyone, including above-average water drinkers. In fact, millions of people drink far more than 2 liters per day. Indeed, some agencies, such as the Food and Nutrition Board (FNB) of the Institute of Medicine, now recommend that males over the age of eighteen drink 3 liters of water per day. 38 Also according to the FNB, over 5 percent of males between the ages of nineteen and fifty consume at least 5 liters of water a day, while 1 percent consume at least 9 liters a day. Over 5 percent of adult females consume over 4 liters a day, while 1 percent of females consume over 5. 5 liters a day. 39
At most, the EPA’s use of the 2-liter consumption figure is designed to protect the average person, not everyone—and certainly not the above-average water drinker.
Mistake six: Not making allowance for fluoride exposure from other sources.
Other sources of fluoride include food and beverages processed with fluoridated water, 40–43 fluoridated dental products, 44, 45 mechanically deboned meat, 46 teas, 47–50 pesticide residues on food, 51 and wine. 52, 53 An estimate of the total ingested fluoride should have been subtracted from the 8 mg/day before proceeding to the calculation of the safe water concentration (MCLG).
People attempting to perform a risk assessment run into another daunting problem at this stage. If they estimate a more appropriate safe daily dose (e. g. , 1 or 2 mg/day), they then find that a conservative estimate for the total fluoride obtained from other sources may already exceed this safe level. How can they then proceed to a non-zero safe level in water? The simple answer is, they cannot. If it is concluded that some people are already getting above the determined safe level from other sources, we cannot allow any fluoride to be added to our water. In other words, one is forced to set the MCLG at zero (as has been done for arsenic and lead). This would rule out the deliberate addition of fluoride to the water and instead shift the whole focus onto the federally enforceable maximum contaminant level (MCL) and the converse issue of how much it would cost some communities with high natural levels of fluoride to reduce those to some compromise level.
Reaction to the MCLG.
Every assumption used by the EPA water division to determine the 1986 MCLG of 4 ppm was inadequate, unscientific, and clearly designed to produce an MCLG that would protect the fluoridation program rather than human health. The manipulations outraged professional employees at the EPA, leading their union to file a “friend of the court” brief in support of the National Resources Defense Fund’s application for an injunction against the 1986 MCLG for fluoride. 54
Focusing on only one of the mistakes discussed above (the choice of end point), the 2006 NRC panel found that the 4 ppm MCLG was not protective of health and recommended that the EPA determine a new MCLG. 55 After four years, as of the time of this writing, the EPA had yet to do that.
Food and Nutrition Board
It came as a considerable shock in 1997 when the Food and Nutrition Board (FNB) of the Institute of Medicine (IOM) announced that it was setting “adequate intakes” (AIs) and “upper tolerance limits” (ULs) for fluoride along with magnesium, calcium, phosphate, and vitamin D. 56 It was a shock because there is no evidence that fluoride is an essential nutrient, as are the other substances appearing on this list, and because its predominant action in fighting tooth decay is topical, not
systemic. 57, 58
In response to a letter sent by about a dozen scientists complaining about this treatment of fluoride as a nutrient, the directors of both the IOM and the National Academy of Sciences made it very clear that they did not consider fluoride a nutrient and instead described it as a “beneficial element” (see chapter 1). 59 However, the FNB proceeded as if it were, indeed, a nutrient.
In determining the UL for fluoride, the FNB, like the EPA ten years before, considered skeletal fluorosis as the only adverse health effect of concern. However, the board dropped the LOAEL to “10 mg/day for 10 or more years. ” While that was clearly an improvement over what the EPA did in 1986, the board then did something quite inexplicable (at least on scientific grounds). In deriving a UL from this LOAEL, it applied a safety margin of 1. 0—meaning there was no uncertainty in applying this so-called safe level to a large segment of the whole population. The terminology the board used was “uncertainty factor” (UF) of 1. 0. 60 Thus it was that the FNB resolved that, for anyone eight years of age or older, it would be safe to consume 10 mg/day for a lifetime. That is a preposterous notion, best explained by the need felt by yet another agency to protect the U. S. water fluoridation program at whatever cost to scientific credibility or the public’s health.
The Fluoridation Forum
We discuss Ireland’s 2002 Fluoridation Forum report in some detail in chapter 24. Suffice it to say here that this panel simply copied the Food and Nutrition Board’s approach of applying a safety factor of 1. 0 to an LOAEL in deriving an upper tolerance level (UL) of 10 mg per day. 61
