Fluoridation

7. Does the fluoridation of drinking water specifically lead to adverse ecological impacts?

The adverse effects of fluoride exposure in humans and the benefits for dental health have been discussed in sections 4.1 and 4.4, respectively and will not be discussed further.

As already indicated in section 3.1, the presence of fluorosilicates in drinking water due to the use of hexafluorosilicic acid or hexafluorosilicate for fluoridation, if any, is very low as fluorosilicates and other species are rapidly hydrolyzed in water to fluoride.

Therefore, this environmental risk assessment will focus only on the fluoride ion.
As indicated in section 3, fluorides occur naturally and are ubiquitous; natural background levels vary with environmental compartments and geological circumstances. Fluorides also enter the environment from human activities besides the fluoridation of drinking water. These can involve the production of aluminium, the production of some building bricks, and the production and use of fertilizers.

Hence SCHER interprets this part of the request as follows: to what extent does the fluoridation of drinking water specifically lead to adverse ecological impacts?

If there were detailed information on exposure and physico-chemical conditions available this approach would therefore consider the extent to which fluoridation adds to the natural background, taking account of regional variations. It should also possibly take account of continental and regional backgrounds that integrate both natural and human sources. It would not consider the extent to which fluoridation might add to other anthropogenic sources at specific sites (e.g. point source emissions from aluminium smelting or diffuse emissions from agricultural use of fertilizers).

The scenario of interest will therefore focus on the environmental exposures arising ofrom the use of fluoridated water as drinking water, personal hygiene, washing clothes and washing dishes. Most of this flows to the environment in drainage water and via sewage treatment works. SCHER did not consider losses due to leakages from water supply pipes and from the use of tap water in irrigation, and therefore soil contamination, since these outputs are not well documented. However, we have focussed on the losses through sewage treatment works. In this route most of the fluorides remain in solution during sewage treatment and pass to the aquatic environment in this way (Walton and Conway 1989). Therefore a negligible amount of fluorides may pass to the terrestrial environment if sludge is spread on land; and/or to atmosphere and land if sludge is subjected to incineration. In the aquatic environment water chemistry will drive distribution between water and sediments. Based on the physico-chemical characteristics of fluoride it is expected that the contamination of soil and the atmosphere are very limited. Fluoride is the most electronegative chemical in the Periodic Table and is highly reactive. Hence in the aquatic environment fluorides are likely to occur as the fluoride anion (Walton and Conway 1989) and therefore this will be the focus of the exposure and effect assessments for the aquatic ecosystems.

To carry out this risk assessment effectively would have required detailed information on ambient exposures and physico-chemical conditions at sites where water is fluoridated. Hence as a pragmatic approach SCHER has assumed further: (1) that the fluoride concentrations in waters used as a source of drinking water reflect local background concentrations; and (2) that those authorities that practice fluoridation would not add fluoride if these background levels exceeded the legally-specified concentrations for fluoride in water for human consumption of 1.5 mg/L in the EU. Hence worst case environmental exposure concentrations will be equal to these legally-specified maxima. On that basis SCHER has used the legally defined concentration for Ireland (0.8 mg/L) and the WHO standard (1.5 mg/L) as appropriate total exposure levels – see section 4.2.1. The value of 3.0 mg/L (scenario 3 in the human health assessment – see section 4.2.1) has not been used in this environmental assessment since this was based on natural concentrations in Finland – i.e. there is no added environmental risk here. Finally, indirect side effects, such as the possible increase in concentrations of lead from the action of fluoride in lead water pipes (section 3.1) are not considered since these scenarios are speculative and difficult to anticipate.

Therefore, SCHER is of the opinion that: 1) fluoride as F- should be considered as the only acting agent; 2) the only source of fluoride in this opinion is the application of fluoride in water supply systems and other sources of fluoride are excluded with respect to potential effects in the environment; 3) as a pragmatic approach it is assumed that the worst-case exposure from fluoridation will be no greater than the allowed legal limits; and 4) the focus of attention for the risk assessment should be the aqueous phase of the aquatic environment.
The physico-chemical properties are mentioned in section 3.2.

Mechanism of action

Fluorides are not essential for most organisms. However, there is evidence that at low concentrations fluorides can enhance the population growth rates of some aquatic algal species (Camargo 2003). Some algae are able to tolerate fluoride levels as high as 200 mg F-/L.

The adverse effects of fluoride on organisms seem to arise from the disruption of key metabolic pathways through the impairment of enzymes, including those involved in nucleic acid synthesis. However, the mechanistic details are as yet unclear.

In fish and invertebrates, fluoride toxicity decreases with increasing calcium and chloride concentrations in the water. The decrease of toxicity with calcium is mainly due to the formation/ precipitation of innocuous complexes such as Ca5(PO4)3F, CaF2 and MgF2. An increase in the concentration of chloride ions might elicit a response in organisms for fluoride excretion. Based on observations in natural media, Camargo (2003) concluded that it should be evident that physiological and genetic adaptation to high fluoride concentrations can occur in wild fish populations.

4.6.3. Aquatic effects

The analysis of the aquatic effects was based on a bibliographic search. From this it appeared that the review of Camargo (2003) covered most of the relevant studies validated by the SCHER. Given the good quality of this review, SCHER has therefore based much of the following analysis of the effects on the information cited in this review. Additional information from field studies (Sigler and Neuhold 1972) did not lead to a conclusive safe level. SCHER concluded that the review of Camargo offered sufficient information of good quality to perform a risk assessment for the environment.

Fish

Freshwater

Acute effects

The most valid data available (96h tests with measured concentration) were reviewed by Camargo (2003) and Metcalfe-Smith et al. (2003). The most sensitive fish was Oncorhynchus mykiss. In worst case soft water conditions (total hardness of 17 mg CaCO3/L) the LC50 96h was 51 mg/L fluoride ion (Camargo 2003).

Chronic effects

Among valid data in the literature, Shi et al. (2009) found the lowest NOEC in fish in 90 days in Acipenser baerii (sturgeon): 4 mg F-/L (measured).

Marine water

Despite of a generally protective effect of chloride ions, Camargo (2003) listed some toxicity data in his review, which were taken as worst case.

Acute effects

Cyprinodon variegatus: LC50 96h more than 500 mg/L (NOEC lethality 500 mg/L).

Chronic effects

Mugil cephalus: NOEC 113d on juvenile development = 5.5 mg/L.

Invertebrates Freshwater

Acute effects

A large number of valid toxicity values in invertebrates at 48h were described in Camargo (2003) and Metcalfe-Smith et al. (2003). The most sensitive species was an amphipod: Hyalella azteca, with an EC50 48h of 14.6 mg F-/L (measured concentrations) with hardness 140–150 mg CaCO3/L (Metcalfe-Smith et al. 2003).

Chronic effects

Metcalfe-Smith et al. (2003) found an IC25 28d of about 4 mg F-/L on Hyalella azteca growth (calculated from the article data on controlled concentration in spiked sediment and overlaying water).

Marine water

Acute effects

Despite the general protective effect of Cl- ions, Camargo (2003) reported some toxicity data, the lowest EC50 96h being 10.5 mg F-/L in the arthropod Mysidopsis bahia.

Chronic effects

Camargo (2003) reported that the female fecundity of Grandidierella lutosa and lignorum estuarine amphipods was shown to be the most sensitive endpoint in a 90 day life-cycle test, with a maximum allowable toxicant concentration (MATC) of 4.15 mg F-/L. It is noticeable that below this value, F- was observed to stimulate female fecundity.

Algae

Freshwater

Acute effects

According to Camargo (2003), among algae species for which growth was not stimulated by fluoride ions, the lowest EC50 96h was shown to be 123 mg F-/L in Selenastrum capricornutum.

Chronic effects

In the same species selection, growth of an algae species Scenedesmus quadricauda with sensitivity generally similar that of Selenastrum capricornutum, was shown not to be inhibited by 50 mg F-/L in 175h. This value can therefore be taken as worst case NOEC for algae.

Marine water

Acute effects

As a general observation marine algal species are less sensitive to fluoride ions. The lowest EC50 96h value of 82 mg F-/L was shown in Skeletonema costatum.

Chronic effects

In the chronic exposure experiments with marine algae cited in Camargo (2003), the lowest tested concentrations of fluoride was 50 mg/L, and the duration was more than 16 days. For algae tested at this concentration, no inhibition was observed. At 100 mg/L, the growth of some species was inhibited, but at most at 30%. Therefore 50 mg/L can be taken as worst case NOEC 72h for algae.

Conclusion on effects

SCHER agreed to use the ecotoxicological data as presented in Table 11 and considered these data sufficiently reliable to be accepted and used in the risk assessment for the environment. From this data set based on the most sensitive taxa, it is evident that freshwater and marine organisms are of similar sensitivity. Therefore, the PNEC for both freshwater and marine water was derived from the whole data set, applying an Assessment Factor (AF) of 10 to the lowest NOEC. (SCHER and its predecessor do not accept the additional safety factor of 10 from freshwater to marine water stated in the TGD). The most sensitive trophic level is the invertebrate one. The chronic toxicity in Hyalella azteca is expressed as IC25 (juvenile growth). As the raw data set is not available in the publication, it is not possible to check if this value is close to the LOEC or NOEC. Therefore the data were not used to avoid excessive uncertainty. The chronic toxicity in Grandidierella sp, very close to the latter value, was used. It is expressed as MATC (female fecundity), from which an NOEC can be derived according to the REACH guidance (MATC/√2).

The PNEC such derived is 0.29 mg/L. However, this value has to be discussed in the light of fluoride ion character as essential oligo-element. Camargo (2003) reported from Connell and Airey (1982) that fluoride concentration below the defined 4.15 mg/L MATC might stimulate Grandidierella sp female fecundity. It is also likely that in most of organisms, fluoride ions stimulate growth and reproduction as essential element. Therefore, using a PNEC such derived has no real meaning, as concentrations below toxic concentrations are considered beneficial. In such a view, Camargo (2003) suggested to use ecologically relevant sensitive endpoints as direct quality levels for safe life in freshwater. Net-spinning caddisfly larvae and upstream-migrating adult salmons, living in soft waters with low ionic content, were found to be the most sensitive organisms, affected by fluoride concentrations higher than 0.5 mg/L. So it is assumed that concentrations lower than this threshold are safe for these extremely sensitive organisms, and therefore for aquatic ecosystems.

Organism	Endpoint	Value (mg/L)
Freshwater
Fish (acute) (Oncorhynchus mykiss)	LC50 (96 h)	51
Invertebrates (acute) (Hyalella azteca)	EC50 (96 h)	14.6
Algae (acute) (Selenastrum capricornutum)	EC50 (96 h)	123
Freshwater
Fish (chronic) (Acipenser baerii)	NOEC (90 d)	4
Invertebrates (chronic) (Hyalella azteca)	EC25 (28 d)	4
Algae (chronic) several species	NOEC (16 d)	50
Marine water
Fish (acute) (Cyprinodon variegatus)	LC50(96 h)	more than 500
Invertebrates (acute) (Mysidopsis bahia)	LC50 (48 h)	10.5
Algae (acute) (Skeletonema costatum)	EC50 (96 h)	82
Marine water
Fish (chronic) (Mugil cephalus)	NOEC (113 d)	5.5
Invertebrates (chronic) (Grandidierella sp.)	NOEC (90 d) = MATC/√2	2.9
Algae (chronic)several species	NOEC (≥16 d)	50
No-effect in both waters	PNEC	0.29

Risk characterization

A simplistic risk characterisation can be carried out by assuming that the fluoridation level is 1 mg/L, that all domestic waters entering sewage treatment works contain fluoride at this level and that most of this flows through the system. This means that worst case fluoride ion concentration in a typical output would be no more than 1 mg/L due to fluoridation – though this will be diluted to a variable extent by rainwater inputs. This means that the effluent would only have to be diluted in receiving water by a factor of at least 3.5 (only 2 if the sensitive species safety threshold is considered) for the fluoride concentration to be reduced below the worst case PNEC of 0.29 for freshwaters– something which seems extremely plausible for most circumstances (default dilution factor taken in the TGD is 10 (TGD 2003). Dilution for effluents entering the marine environment would have to be greater; but again that seems plausible (the default dilution factor taken in TGD for marine ecosystems is 100 (TGD 2003)).

The only detailed work that has been carried out on the consequences of fluoridation of drinking water for concentrations of F in sewage treatment effluents was done by Osterman (1990) and supports the conclusion from the simplistic assessment. This paper presents a mass balance approach to develop a series of mathematical equations that describe the fate of fluoride added to drinking water in a typical municipal water management system. The ionic mass of fluoride entering the aquatic system from all sources was calculated, its distribution followed and its fate examined. The city of Montreal in Canada was used as an example but it is SCHER's view that this approach can be applied broadly. In this system fluoride was added to obtain levels between 0.7 and 1.2 mg/L. Based on the fluoridation level and the characteristics of the water supply situation in Montreal, the estimated daily average fluoride concentration at less than 1km distance from the effluent outfall was 0.22 to 0.34 mg/L. If this is compared with the safe threshold of 0.5 mg/L, no unacceptable risk for aquatic organisms is expected.

Clearly this study is focused on a particular site. To check the generality of the results, SCHER further has carried out an analysis using the European Union System for the Evaluation of Substances (EUSES) (EC 2004).

SCHER recognizes that this model has been designed to be applied for organic and hydrophobic substances in the framework of new and existing substances and biocides (EC 2004) but is of the view that treated cautiously the model can give further insight into the likely consequences of fluoride for aquatic systems.
The addition of fluoride to drinking water is analogous to the addition of disinfectants to drinking water and this version of EUSES has been adopted in the following analyses.

In addition it should be kept in mind that the scenarios included in EUSES are conservative.

The following assumptions have been adopted by SCHER:
1. addition of fluoride according to PT5 in analogy to drinking water disinfection;
2. the dose applied is 0.8 (normal dose) and 1.5 mg/L, based on the Council Directive 98/83/EC of 3 November 1998 on the quality of water intended for human consumption (see section 4.2.1, human part);
3. thephysico-chemicalcharacteristicsareasindicatedinTable1;
4. the effect data are as indicated in Table 11. The following 2 cases are presented:
1. Case 1: a dose of 0.8 mg F-/L as the normal dose for fluoridation of drinking water,
2. Case 2:adoseof1.5mgF-/L, based on the reference dose of WHO (2006), The main results of the calculation of the risk characterisation ratios (RCR), defined as the ratio between the Predicted Environmental Concentration (PEC) and the Predicted No-Effect Concentration (PNEC) are that case 1 leads to an RCR of 0.276 and case 2 to an RCR of 0.517 (see Appendix II).

From these different lines of evidence, SCHER deduces that fluoridation of drinking water will not result in unacceptable effects to the environment as RCR-values are below 1.

Conclusions

Based on three lines of evidence, a simplistic risk assessment, mass balance modelling and a modified EUSES analysis, SCHER is of the opinion that adding fluoride to drinking water at concentrations between 0.8 mg F-/L and the reference dose level of WHO (1.5 mg F-/L) does not result in unacceptable risk to water organisms. Due to the electronegativity of the F ion SCHER is of the view that there will be little partition to solids in the sewage treatment process.
It follows that sewage sludge is unlikely to become contaminated and, in turn, this means that the contamination of soils and terrestrial systems is unlikely from this source. There is still the possibility of direct soil contamination from leakage from the water supply system and by irrigation using tap water. SCHER was not able to carry out risk assessments here due to lack of exposure data. If there were the possibility of significant exposures in particular sites from these sources then more work would be necessary to asses risk to the soil ecosystem. Atmospheric releases from the incineration of sewage sludge are unlikely.