The report from the EU Ad Hoc Working Group on dental amalgam: a Critique

About this document

This document is a response to a draft version of a document issued by the “EU ad hoc working group on dental amalgam”. It is not complete, but as regards the aspects dealing with the systemic health risk from dental amalgam, it should cover the most important points.

All numbered sections refer directly to the corresponding sections in the working group’s document.


The working group’s draft report is a surprisingly biased and incomplete document. The 1994 report of the Swedish National Board of Health and Welfare is also biased, but at least that group made sure that they were up to date on the scientific literature on amalgam.

The working group is clearly not up to date on the literature, nor has it done what is necessary to update itself. In nearly all the significant areas of research relevant to the health risk from amalgam, most of the interesting recent research has been omitted. This is true of amalgam exposure estimates, occupational mercury exposure, heart disease, Alzheimer’s disease, Parkinson’s disease and reproductive toxicity. This sorry state of affairs could have been avoided by doing a few searches of the most well-known medical citation database, Medline, but apparently the working group was either unwilling or unable to do such searches. This is extremely odd, as literally anyone can do it via the Internet or at any medical library with a little help from a librarian.

As a consequence, the draft report is mostly a review of the 1994 state of knowledge, with a few bits of more recent science thrown in. The selection of more recent science is clearly biased, and the studies that are cited are described in a biased manner. Criticism is directed preferentially at studies that seem to indicate health risks from amalgam, whereas flawed studies that seem to support the safety of amalgam are cited without comment.

The working group seems to have only a rudimentary knowledge of the toxicology of mercury, as evidenced by their apparent belief that a lack of correlation between blood mercury and amalgams indicates that exposure from amalgams is not signficant. In this, they ignore the fact that their own trusted sources say that blood mercury is not a reliable measure of low-level mercury exposure.

A lack of understanding of the fundamental principles of risk assessment is in evidence in chapter 7. The working group has failed to apply the risk assessment procedures specified in the toxicology textbooks they themselves claim to have used for risk assessment. For this reason, their discussion of toxicological risk from amalgam is off by a factor of 100.

Risk management perspectives

If public bodies are to make good decisions about regulating potential hazards, citizens must be well-informed. The alternative of entrusting policy to panels of experts working behind closed doors has proved a failure, both because the resulting policy may ignore important social considerations and because it may prove impossible to implement in the face of grass-roots resistance. (Granger Morgan 1993)

On this, the EU ad-hoc working group on dental amalgam is one of the most obvious examples of bad risk management. If there ever was an example of “experts working behind closed doors”, this is it. (Except for the question of whether the working group members really are experts on the risk from dental amalgam.)

Also, the risks and benefits are not necessarily of the same kind, which means that weighing risks and benefits becomes a value judgement. Is the risk of systemic toxicity from amalgam worth the (possible) cost savings from using it? This depends on how much value one puts on health. Such value judgements are political, not scientific issues, and should not be decided by expert groups.

One of the working group’s own sources, the 1993 USPHS report, presents the following Recommended PHS Strategy for education:

Develop a public and professional educational campaign to explain to dental personnel and consumers what is and is not known about the safety of dental amalgam. In order to be credible, the program must address both the benefits of amalgam and the controversy that exists regarding the possible biological effects.
Lead agency: CDC, jointly with FDA and NIH.

So in 1993, the USPHS decided that the public needed be informed of the scientific controversy surrounding amalgam. In 1997, the EU Commision appoints a working group that clearly contains representatives of only one side of the controversy. Thus, the way this has been handled has been deceptive from the very outset.

5.2.1 Inhalation of mercury vapor

Conspicuously missing from table 1 is the Weiner & Nylander estimate of 4-19 µg per day from 1995.

Richardson’s estimate based on urinary mercury is cited, but not M. Allan’s estimate based on Olsson & Bergman’s method. (Richardson 1995, Richardson & Allan 1997) Richardson gives the Allan estimate as 3.74 µg per day.

5.2.4 Mercury in blood

The working group writes:

Fish consumption, however, rather than the presence of dental amalgam fillings was found by Möller-Madsen et al (1988) and Åkeson et al to be more significantly associated with whole blood mercury levels.

Although it is not explicit, this may be interpreted as indicating that fish is the predominant source of mercury exposure. However, this is an invalid inference, since blood is unsuitable as an indicator of low-level inorganic mercury exposure as confirmed by the 1993 USPHS report (page III-4) and by the 1991 WHO report (page 61). The working group’s seeming uncritical acceptance of the Stoz et al. study (5.3.5) as evidence that fetal exposure to amalgam mercury is not a problem seems to indicate that this is the intended meaning, not a misinterpretation.

5.3.2 Neurological, neuromuscular and cardiovascular disease

No mention is made of the Finnish study (Salonen et al. 1995) that found a doubled risk of coronary heart disease in men with high fish consumption and high mercury levels in hair. The mechanism indicated by this study is lipid peroxidation caused by mercury (measures of lipid peroxidation were also correlated with mercury). Since lipid peroxidation is a phenomenon which is primarily known to occur with inorganic mercury, there is reason to assume that amalgam mercury has the same effect.

The working group states that “currently the weight of the evidence is that there is no link between Alzheimer’s disease and mercury”. To support this, it cites two sources. One does not appear to be a peer-reviewed scientific document, the other (Saxe et al.) does not purport to be a study of Alzheimer’s disease. It is a study of cognitive function in older nuns, age 75-102. Since the study includes no Alzheimer’s patients, it is unclear whether the results have relevance to Alzheimer’s disease.

The working group also ignores the animal evidence that shows that mercury at levels similar to those from amalgam has the ability to induce Alzheimer-like changes in brain tissue (Duhr et al. 1991, Palkiewicz et al.1994, Pendergrass & Haley 1997).

Regarding Parkinson’s disease, the group fails to mention two case-control studies studies suggesting a link between Parkinson’s disease and mercury. In a case-control study in Singapore, Ngim & Devathasan (1989) found a clear dose-response relationship between mercury levels in blood and Parkinson’s disease. In a recent German case-control study (Seidler et al. 1996), the patients reported a significantly larger number of amalgam-filled teeth before their illness than control subjects.

On heart disease, the working group states:

In a Swedish epidemiological study (Ahlqwist et al 1993), 1462 women were followed for over 20 years. No correlation between dental amalgam fillings and cardiovascular disease, diabetes cancer nor early death was found. In the same population dental amalgam fillings were found not to be associated with the impairment of kidney function or immunological status.

This paragraph actually refers to three papers by Ahlqwist et al. (1988, 1993 and 1995.) Relvant to this research is the following quotation from one of the working group’s own trusted sources, the 1993 USPHS report:

Part of the reason for the dearth of information on whether there are health effects from the mercury in amalgam is that the few human studies that have investigated this issue have been too small or flawed in design to detect an effect. (Page 3)

Since this was published in 1993, it applies at least to the 1988 Ahlqwist et al. paper. So this study can be assumed to be too small, too flawed, or both. Friberg (1995) expands on the issue of size:

Furthermore, despite negative results in epidemiological studies, the statistical power is not high enough to exclude the occurrence of effects in a few percent of the population at still lower exposure levels. As amalgam is used so widely already, an effect in a few percent of the exposed population would mean that very large population groups could be affected.

Richardson (1995) expands on the issue of methodological flaws:

With regard to the epidemiological literature, there have been no adequately conducted epidemiological studies of amalgam bearers, with proper controls and objectively measured signs and symptoms. The studies which have been reported (Ahlqwist et al. 1988, 1993; Lavstedt and Sundberg, 1989) fail to provide unequivocal evidence of absence of effects due largely to methodological weaknesses. Lack of adequate control groups, potential bias in subjectively reported symptoms, and failure to focus on disease states most likely to arise from amalgam or Hg exposure limit their value in the current assessment.

When the working group puts the Ahlqwist et al. cohort at 1462 people, this is consistent with some of the abstracts, but in reality it is misleading. The number of people actually used for the statistical comparisons were 653 (1988, 1995) and 812 (1993).

The 1993 study, although heart disease is one of the subjects of its investigation, fails to control for even the most obvious risk factors such as smoking. This is in contrast to Salonen et al., whose study is extremely well-controlled.

It is unclear how good the Ahlqwist et al. exposure measurements are. From their cohort of middle-aged women, they selected a “low-exposed” group with 4 or fewer amalgam fillings. In this age group, it is highly unlikely that most of these had perfect or near-perfect teeth. It is more likely that they had other restorations, including perhaps gold crowns with amalgam underneath.

To cite Ahlqwist et al. without mentioning any of these shortcomings is biased reporting.

The working group also quotes from the Swedish 1994 report as follows:

Also that scientific studies show that dental amalgam does not contribute to cardivascular disease in women.

This claim does occur in the report summary, but it is false or misleading, since the report text cites only one study, rather than “studies”. (Ahlqwist et al 1993)

5.3.4 Nephrotoxicity

The Eti et al. study is cited as follows:

Eti et al (1995) reported on the number of dental amalgam restorations or surfaces and various markers such as lysosomal enzyme, urinary mercury and proteinuria levels. They found no correlation between urinary mercury levels and the number of fillings or enzyme excretion and concluded that the small differences detected in urinary enzyme level excretion were insufficient to indicate renal injury.

What this paragraph fails to state is the most noteworthy finding in this study: that “the small differences detected in urinary enzyme level excretion” were positively correlated with the number of amalgam fillings! It is as if the working group is reluctant to report in plaintext any finding that might throw doubt on the safety of amalgam. (Although, in all fairness, in section 7.4.4, this study is cited in a way that makes it possible to infer what the actual finding was.)

It is true that the differences found indicate only a “slight effect”, but it is worth remembering that this was a relatively small study with an average age of 30. Discrepancies much larger than the average difference were found in single individuals, and older people would probably be more susceptible to nephrotoxic effects of mercury, as is known to be the case with cadmium (Elinder & Järup 1996).

5.3.5 Reproductive toxicity, fetotoxicity and fertility

The working group’s treatment of this subject is extremely biased and incomplete. The working group first cites the Drasch et al. study that found a correlation between on the one hand mercury in organs of newborns and on the other the mother’s amalgam fillings. This is accompanied by some critical comments from the working group. Then they cite Stoz et al. who found that blood mercury in both the women and the children were poorly correlated with the mother’s amalgams. Stoz et al.’ s conclusions are quoted without comment as if this study had repudiated the Drasch et al. findings. However, as Drasch points out in a reply (Drasch 1995), what Stoz et al. have in fact done is:

  1. To confirm earlier findings that blood is a poor measure of amalgam mercury and that the number of fillings correlates better with the mercury in organs. The fact that blood is unsuitable as an indicator of low-level inorganic mercury exposure is confirmed by the 1993 USPHS report (page III-4) and by the 1991 WHO report (page 61).
  2. In spite of their own postulate that the placenta protects the child against mercury, they have demonstrated the exact opposite, since there was an almost perfect correlation between the mother’s and the child’s blood mercury.
In addition to its biased reporting of the Drasch/Stoz discussion, the working group ignores a number of studies that support the Drasch et al. results. Most of these are too recent to be covered by the earlier reviews.

Transfer of amalgam mercury across the placenta and concentration in fetal tissues has been demonstrated in sheep (Vimy et al. 1990). Several animal studies have shown fetal injury or development disturbances from mercury vapor exposure during pregnancy (Danielsson et al 1993, Warfvinge et al. 1994, Söderström et al. 1995, Fredriksson et al. 1996). In at least one case (Söderström et al. 1995) this occurred at exposure levels causing tissue concentrations similar to those found by Drasch et al. in human newborns.

The correlation between amalgam fillings and mercury in breast milk has been documented in several studies of humans (Oskarsson et al. 1995, 1996, Vimy et al. 1997) and animals (Yoshida et al. 1994, Vimy et al 1997). Oskarsson et al. (1996) find that the exposure of the infant via breast milk corresponds to approximately one half of the tolerable daily intake for adults recommended by the WHO. The authors conclude that efforts should be made to decrease the mercury burden in fertile women.

The working group’s portrayal of the WHO report’s review of reproductive toxicity is about as biased as it is possible to make it. When discussing a Polish study that showed a correlation between occupational mercury exposure and abortions / malformations (misleadingly rendered by the working group as “fertility”), they attach critical comments, and mention that “Larsson (1995) expressed major reservations about the Polish study”. When referring to “four reports which did not find any link between occupational mercury exposure and the fertility of female dental staff”, they fail to mention the WHO’s comments about the limitations of these studies. Four additional studies cited by the WHO that reported correlations between mercury and reproductive disturbances are simply ignored by the working group. Other general health complaints attributed to dental amalgam.

The working group states that “a number of authors have found a psychogenic component behind such complaints”, without mentioning that none of these studies have a methodology that is suitable for demonstrating a cause-effect relationship. Among the most common problems in these studies are:

  1. Uncritically inferring causation from correlation. The mere existence of mental problems in patients who attribute their health problems to amalgams is used as evidence that the psychological problems are the cause, in spite of the fact that mercury toxicity is known to produce similar psychological symptoms, and that perfectly normal people frequently become depressed when physically ill.
  2. Arbitrary labeling of symptoms as “psychological” and using this as evidence of psychogenic etiology. For instance, Hugoson (1986) lists tremor as a “psychological symptom” in spite of the fact that it is one of the most characteristic symptoms of mercury poisoning.
  3. No clear specification of hypotheses being tested.
In a review of studies of multiple chemical sensitivities syndrome and pscyhogenic origins, Davidoff & Fogarty (1994) listed 15 common methodological errors and found that the average study had 9.6 of these problems. The studies on amalgam are probably even more dubious. All studies reviewed by Davidoff & Fogarty at least had a control group. Of the 12 studies reviewed by the Swedish expert group on “psychiatric aspects” of amalgam, only 5 used a control group.

The working group then quotes the Swedish 1994 report on the effect of amalgam removal:

Follow up of patients at medical centres showed that many, in some cases the majority of subjects, had persistent symptoms and discomfort, which were generally attributed to other causes and which remained even after the removal of amalgam fillings.

This is based on biased and poor choice of data by the Swedish expert group. They have arbitrarily chosen to restrict themselves to material collected at Swedish medical centers instead of selecting data for quality and reliablity. The problem with most of the expert group’s data is that they are not reports of systematic clinical trials, but rather observations of patients that have had amalgams removed on their own initiative. This means a total lack of control of removal procedure, speed of removal, and to some extent time since removal. Studies that report on more controlled removal are not mentioned by the Swedish expert group (Lichtenberg 1993, Zamm 1990, Godfrey & Campbell 1994, Redhe 1991). These show better than 90% improvement rates. Nor do the Swedish expert group cite other, less well-controlled studies that have also demonstrated high improvement rates (Hanson 1986, Siblerud 1990, Östlin 1991).

Of the studies the Swedish expert group did cite, one is the ambiguous study mentioned by the EU working group . In another one (Lindfors et al. 1994), the patients were selected from a group who sought medical attention after they had had their amalgams replaced. So these were presumably the ones who were dissatisfied with the results of removal. The third is misquoted in the Swedish report as follows: “it has been shown that the patients’ so-called illness quotient (number of days on sick-leave or invalid pension) does not decrease after removal of amalgam fillings.” The report itself states that “with respect to the great differences in the number of days on sick-leave that one or two people can cause, no general conclusions can be drawn from this study.” The only study quoted by the Swedish expert group that can reasonably be said to support the idea that amalgam removal does not have much of an effect is Bjerner & Hjelm (1994), although even this study is problematic.

Another recent study (Lindqvist & Mornstad 1996) showed improvement in objectively measured immune parameters after amalgam removal.

None of the studies allow a definite conclusion that the patients’ improvement was due to amalgam removal, but the best data we do have available show very high improvement rates (70% or more) almost without exception.

The working group’s cites the 1996 Berglund and Molin study that found no difference in mercury exposure between a group with self-diagnosed amalgam-related illness and a control group. However, a key piece of information is missing. It is also missing from the study’s abstract. I quote:

The criteria for participating in the control group…were that they had to match the symptom group regarding age and number of amalgam surfaces [my emphasis].

Mercury exposure is known to be related to the number of amalgam surfaces (in fact, it is perhaps the most commonly used measure of amalgam mercury exposure in epidemological studies). So matching the two groups in this way will obviously tend to make them more similar in exposure. So, from the experimental design alone, the finding of no significant difference between the groups should not be a surprise. Using an experimental design that tends to reduce or even remove the effect one is looking for is a very dubious way of conducting research. It is obvious that this study is of no value in assessing whether people with self-diagnosed amalgam-related illness have a greater mercury exposure than healthy people with amalgams.

7.3 Dental amalgam

The working group states:

An initial observation of the group was that there was a limited amount of reported clinical material identified in chapter 5, in relation to toxicity due to dental amalgam fillings in the general population, even though dental amalgam has been placed in billions of teeth. However it was acknowledged that whilst this did not indicate safety or a lack of risk, it may provide an estimate of the magnitude of the risks posed by dental amalgam since, given the extent of the exposure in the population, significant risks should have been observable.

Here, the working group seems to have gotten lost in its own rhetoric. The risk of lung cancer and heart disease from smoking is not “observable” except by epidemiological studies, even though the risk is high and a large proportion of the population is exposed. In fact, “risk”, and the cause-effect relationship on which it is based, is an abstraction that cannot be observed directly, only inferred. Sick people are observable, the etiology of their disease is not. Also, exposing nearly all members of a population makes the risk more difficult to ascertain, not less, since the risk will be most clearly evident in comparing exposed and non-exposed groups.

7.4.2 Exposure

The working group’s “risk assessment” suffers from the following shortcomings.

  • They appear to be unaware of the recent evidence on occupational exposure.
  • They have used exposure estimates that are likely to be too low.
  • They have not considered the effects of individual variability.
  • They have not established an acceptable mercury intake.

No recent evidence

The working group quotes the 1991 WHO report as a source on studies for low occupational exposures:

The US NIH report suggested a value of 50 µg/m3 as the cut-off for subclinical effects but the 1991 WHO report includes some studies which report subclinical effects at estimated occupational exposures between 30 and 50 µg/m3.
Those studies considered in detail (for example Fawer et al. 1983) suggest these lower exposure estimates are unreliable (i.e. using a single measurement taken under the controls existing at the time of testing as the basis for historical exposure). For the purposes of this comparison, however, it was felt that the WHO range would provide a conservative estimate of the range of occupational exposures associated with subclinical effects (i.e. 30 to 50 µg/m3).

In fact Fawer et al. found evidence of toxicity at 26 µg/m3, and several other studies, most of them post-1991, confirm this. The working group does not appear to be aware of this evidence which indicates that 30 to 50 µg/m3 is not a conservative estimate but rather the opposite (Soleo et al. 1990, Ngim et al. 1992, Liang et al. 1993, Gonzales-Ramirez et al. 1994, Echeverria et al. 1995).

Low exposure estimates

The low-end and most likely incorrect exposure estimates are used.

The majority of the estimates are in the 1-5 mg per day range with two results in the 15 - 20 mg per day range.

The “majority” can only be acheived by quoting each recalculation as a separate estimate. However, this is not the key issue. It is more important to use the most recent and best estimates than to use majority voting on old and new alike. The working group has used reevaluations of exposure from around 1990 uncritically without realizing that later work has superseded these.

The Weiner & Nylander paper from 1995, which the group has failed to discover, demonstrates that the lowest estimates must be in error. Based on this paper, it is clear that

  1. urinary excretion of mercury in Swedish subjects is approximately 3.1 µg per day
  2. the contribution from non-amalgam sources is approximately 0.4 µg per day
So urinary excretion of amalgam mercury alone is 2.7 µg per day, which is higher than the Olsson & Bergman estimate of 2 µg per day. Furthermore, mercury vapor is also eliminated by other pathways, leading to the estimate of 4–19 µg per day. The working group does cite the Jokstad et al. paper from 1992, which arrives at a similar estimate (10–12 µg per day) by a similar methodology.

Individual variability not considered

Individual variability in the number of fillings or in mercury uptake is not considered. The fact that many people have higher-than average exposures makes the risk higher for these, who of course constitute a large proportion of the population.

Also not mentioned is the evidence that amalgam mercury exposure in some cases is much higher than the figures quoted so far, sometimes even exceeding occupational exposure limits in people with bruxism and in heavy gum chewers (Barregård et al 1994, Sällsten et al. 1996). The 1994 Swedish report acknowledges the validity of this evidence, but avoids adressing its full significance by failing to discuss the neurological implications of this exposure. Instead, the report focuses only on nephrotoxicity in relation to these high exposures.

The working group says:

Users may be potentially exposed to greater mercury exposure than this. However, provided the guidance on handling mercury and dental amalgam in the information provided by the manufacturer and agreed codes of practice are followed in line with national legislation establisheing acceptable occupational exposure standards, such risks are manageable and can be controlled.

This is quite an interesting statement. The mention of occupational exposure standards seems to imply the possibility that amalgam exposure can exceed such standards. This suggests that the working group is aware of evidence that this can happen. Otherwise, why would they even mention it? But if they know, and are deliberately not mentioning it, that would be deliberate deception. Of course, since this is a draft document, it is possible that they planned to discuss the evidence but forgot to do so.

No acceptable mercury intake established

Toxicological risk assessment uses uncertainty factors to derive acceptable intakes. The working group does not apply this method, nor do they acknowledge its existence, nor do they give any indication that they understand why the nature of a toxicological dose-response relationship makes it necessary.

The working group’s own sources confirm this, but apparently the working group has not been able to perform a risk assessment according to the instructions in their own toxicology textbooks. They cite three toxicology texts, Cassarett and Doull’s Toxicology, Hayes: Principles and Methods of toxicology and Ballantyne et al: General and Applied Toxicology.

General and Applied Toxicology has this to say about risk assessment:

It is the objective of risk assessment processes to assess the probability that adverse health effects will develop from known, or suspected, toxic xenobiotics in the environment (e.g. drinking water or air) or workplace. Such quantitative risk assessments are most frequently conducted for worktime or lifetime exposure to low concentrations of xenobiotic. They are based on extrapolating dose-response relationships from animal studies, or occasionally human epdiemiological data (1) to determine risk at known or anticipated range of occupational or environmental exposure doses, or (2) to assess “risk- free” doses. (Page 33.)

So it is clear that risk assessment is based on extrapolation, but this text gives no further details on how to extrapolate. Fortunately, Principles and Methods of Toxicology does.

The general approach for setting exposure limits for systemic toxicants thus differs from that commonly used for carcinogens. When animal studies are used to derive limits, uncertainty factors are applied to the dose levels in the animal experiments to account for the lack of information on how to extrapolate from the animal species to humans. While these factors have traditionally been termed “safety” factors, “uncertainty” factors more accurately describes their use, which is to account for uncertainties in the relationship between exposure to a chemical in an animal study and a particular effect, and in the relationship between lifetime daily exposure to the same chemical in the general population of humans and the likelihood of a particular effect. The application of the appropriate uncertainty factors to an experimental exposure yields a level defined as the acceptable daily intake (ADI), which represents a daily intake level of a chemical in humans which is associated with no risk of adverse effects.
Recently, the EPA has decided to refer to such an exposure level as the risk reference dose or RfD. The basis for the change in terminology is that the ADI does not represent a magic dividing line between safe and nonsafe but is an exposure level, derived through a consistent methodology, at which the chance of adverse effects is low. It is still possible that there are some humans who, because of special susceptibility, could suffer an adverse effect at the RfD level. (Page 18)

(The first paragraph may be misinterpreted to mean that extrapolation using uncertainty factors is used only for animal data. This impression is incorrect, as evident from a later statement that one of the factors would be applied to human studies, and from the fact that only one of the factors is for extrapolation from animals to humans.)

Then the text goes on to specify the factors, a 10-fold uncertainty factor for each of the following: 1) variations in susceptibility in humans, 2) extrapolation from animal to human data, 3) extrapolation from subchronic to chronic exposure and 4) extrapolation from a LOAEL (lowest adverse effect level) to a NOAEL (no observed adverse effect level). Two of these factors (1 and 4) apply when extrapolating from occupational studies to amalgam exposure, and so the correct uncertainty factor for this case is 100. Since the working group has failed to apply any uncertainty factors, their discussion of toxicological risk from amalgam is off by a factor of 100.

Cassarett and Doull’s Toxicology also discusses risk assessment, but the risk assessment chapter is new in the fifth edition. The fact that the working group has used the second edition, which is now 16 years obsolete (!), may help explain their lack of insight. The new edition is crystal clear about the need for extrapolation:

…the risk assessor is normally interested in low environmental exposures of humans, which are way below the experimentally observable range of responses in animal assays. Thus low-dose extrapolation and animal-to-human risk extrapolation methods are required and constitute major aspects of dose-response assessment.” (Page 80.)

The only true risk assessment of dental amalgam that has been published in the peer- reviewed scientific literature is that by Richardson & Allan (1996). The working group appears to be unaware of this document’s existence or else finds it too embarrasing. Richardson & Allan calculate a Tolerable Daily Intake (TDI, another name for ADI) for mercury vapor. They find that it corresponds to two (2) or (4) fillings using either of two different exposure calculation models. They also compare the two models to exposure limits from risk assessments by the ATSDR (Agency for Toxic Substances and Disease Registry a branch of the US Public Health service) and the USEPA. This works out to (0) mercury fillings for the USPHS Standard, nine (9) and eleven (11) mercury fillings for the USEPA Standard. Richardson and Allan conclude:

It is apparent, therefore, that there is uncertainty in what constitutes a level of Hg vapor exposure that is acceptable, tolerable, or of minimal risk. Subsequently, there is uncertainty in the precise number of amalgam-filled teeth that can be considered acceptable, tolerable, or of minimal risk. However, it is also apparent that the continued unconditional and unlimited use of amalgam as a dental restorative material, the placing of up to 25 amalgam fillings in one individual, is not supported by the available risk information.

Since the working group, like most people, does not understand the principles of risk assessment, it is worth explaining why uncertainty factors are used. The working group bases its comparison on a LOAEL (lowest observed adverse effect level) of 30 to 50 mg mercury per m3 air. The key point to understand is that a LOAEL is higher than a NOAEL (no observed adverse effect level), and that even a no observed adverse effect level is higher than the no adverse effect level. The fact that an effect cannot be observed does not mean that it does not exist, a fact that is routinely acknowledged in risk assessment. The reason is that studies of animals and occupational exposure are limited in their sensitivity. Usually, toxicity that affects less than 10% of the population cannot be detected (Clarkson 1991). So a NOAEL can easily be an exposure that has a toxic effect on 5–10% of the population. At lower levels, fewer individuals will be affected, but it will not be close to zero until we reach the true population threshold, an actual no adverse effect level. This threshold cannot be measured directly, so it must be extrapolated from a NOAEL or a NOAEL. This is what uncertainty factors are used for: to extrapolate in an attempt to find a level that is truly safe.

(A similar illustration can be found in Clarkson 1991)

This kind of methodology is routinely used in assessing risks from drugs, environmental chemicals and food additives. In contrast, the working group appears to believe that any exposure below the LOAEL is safe, and that it only gets safer below that level. This assumption is equivalent to the belief that the dose-response curve suddenly and inexplicably drops to zero below the LOAEL.

The working group says:

…This suggests that actual exposure to mercury from dental amalgam is 35 to 285 times lower than the subclinical effect level…This could be considered a sufficient margin of safety even in the absence of any benefits, however dental amalgam also has a number of advantages (section 7.6). It was concluded that the toxicological risk was ‘broadly acceptable’ that is in the lowest category of risk according to the concepts outlined in Annex 3 of this report.

If the working group had followed the instructions in their own textbooks, they would have applied an uncertainty factor of 100 (as Richardson & Allan did). This would mean that an exposure 35 times below the LOAEL would be approximately 3 times the acceptable daily intake. The working group owes us an explanation of why they think a toxicological risk that would normally be considered unacceptable even in the presence of benefits (for instance, the benefits of cars or food preservatives) is acceptable even in the absence of benefits where dental materials are concerned.

7.4.5 Reproductive toxicity, fetotoxicity and fertility

This section suffers from the working group’s lack of up-to-date knowledge of the scientific literature, as detailed in my comments on section 5.3.5. Drasch’s study is dismissed without consideration of the supporting evidence, nor of the failure of the Stoz et al. study as a “rebuttal”.

The working group cites a NOAEL of “at least 100 µg/m3…for embryotoxic and teratogenic effects” without including a reference for this claim. It is unclear whether this refers to continuous exposure during the entire pregnancy, or just a single or a few days of exposure, as is commonly used in such studies. In any case, a risk assessment of reproductive toxicity based on animal studies without a review of the most recent data is meaningless.

The 1996 US Food Quality Protection act demands an addtional uncertainty factor of 10 to protect children pre- and post-natally in the absence of reliable data on the effects of exposure in these groups. Applying this to mercury would bring the ADI for children far below the exposures they get from their mother’s amalgams via breast milk, and the ADI for the mothers down to a level that would require zero fillings.

7.5 Alternatives to dental amalgam

The working group is of course justified in pointing out that a detailed risk assessment of alternative materials requires more information about the various substances released. However, the working group’s comments are biased in at least two ways:

  1. They neglect to mention the vast difference in systemic toxicity between mercury and the organic substances released from composites. For just one example, the NOAEL for local irritation from methylmethacrylic acid is 104 mg/m3 in chronic rat studies (Greim et al 1995). This level is 2000-3000 times higher than the working group’s postulated LOAEL for occupational exposure to mercury. In fact, mercury causes acute death by asphyxiation at levels that are lower than the chronic irritation levels for the acrylates and methacrylates (27 mg/m3, Livardjani et al. 1991).
  2. They hypothesize “a greater incidence of hypersensitivity to alternative materials than to dental amalgam” based on “users [dentists, production workers?] of these materials”, without mentioning that users of mercury know its high toxicity and nowadays know how to protect themselves, whereas acrylic resins have only recently started to become the subject of similar caution.


Ahlqwist M, Bengtsson C, Furunes B, Hollender L, Lapidus L: Number of amalgam tooth fillings in relation to subjectively experienced symptoms in a study of Swedish women. Community Dentistry & Oral Epidemiology 1988; 16 (4): 227-31

Ahlqwist M, Bengtsson C, Lapidus L: Number of amalgam fillings in relation to cardiovascular disease, diabetes, cancer and early death in Swedish women. Community Dentistry & Oral Epidemiology 1993; 21 (1): 40-4

Ahlqwist M, Bengtsson C, Lapidus L, Lindstedt G, Lissner L: Concentrations of blood, serum and urine components in relation to number of amalgam tooth fillings in Swedish women. Community Dentistry & Oral Epidemiology 1995; 23 (4): 217-21

Barregard L, Sallsten G, Jarvholm B: People with high mercury uptake from their own dental amalgam fillings. Occupational & Environmental Medicine 1995; 52 (2): 124-8

Clarkson TW: Principles of risk assessment. [Review]. Advances in Dental Research 1992; 6: 22-7

Danielsson BR, Fredriksson A, Dahlgren L, Gardlund AT, Olsson L, Dencker L, Archer T: Behavioural effects of prenatal metallic mercury inhalation exposure in rats. Neurotoxicology & Teratology 1993; 15 (6): 391-6

Davidoff AL, Fogarty L: Psychogenic origins of multiple chemical sensitivities syndrome: a critical review of the research literature. Arch Environ Health 1994; 49 (5): 316-25

Drasch G: [Comment on F. Stoz et al.: is general amalgam prohibition justified? (letter; comment)]. [German]. Zeitschrift fur Geburtshilfe und Neonatologie 1995; 199 (4): 176-7

Duhr EF, Pendergrass JC, Slevin JT, Haley BE: Hg2+ induces GTP-tubulin interactions in rat brain similar to those observed in Alzheimer's disease. FASEB Journal 1991; 5: A456.

Echeverria D, Heyer NJ, Martin MD, Naleway CA, Woods JS, Bittner AC Jr: Behavioral effects of low-level exposure to elemental Hg among dentists. Neurotoxicology & Teratology 1995; 17 (2): 161-8

Friberg LT. Concluding remarks. In: Friberg LT, Schrauzer GN. Status Quo and perspectives of amalgam and other dental materials. International symposium proceedings. G. Thieme Verlag Stuttgart, 1995: 134-136.

Godfrey M, Campbell N: Confirmation of mercury retention and toxicity using 2,3-dimercapto-propane-sulphonic acid sodium salt (DMPS). Journal of Advancement in Medicine 1994; 7: 19-30.)

Granger Morgan M: Risk analysis and management. Scientific American 1993; 269(1): 24-30

Greim H, Ahlers J, Bias R, Broecker B, Hollander H, Gelbke HP, Jacobi S, Klimisch HJ, Mangelsdorf I, Mayr W, et al: Assessment of structurally related chemicals: toxicity and ecotoxicity of acrylic acid and acrylic acid alkyl esters (acrylates), methacrylic acid and methacrylic acid alkyl esters (methacrylates). [Review]. Chemosphere 1995; 31 (2): 2637-59

Hanson M. Changes in health caused by exchange of toxic metallic restorations. Bio-Probe Newsletter 1989; March:3-6.

Herrstrom P, Hogstedt B: Clinical study of oral galvanism: no evidence of toxic mercury exposure but anxiety disorder an important background factor. Scandinavian Journal of Dental Research 1993; 101 (4): 232-7

Liang YX, Sun RK, Sun Y, Chen ZQ, Li LH: Psychological effects of low exposure to mercury vapor: application of a computer-administered neurobehavioral evaluation system. Environmental Research 1993; 60 (2): 320-7

Lichtenberg H: Elimination of symptoms by removal of dental amalgam from mercury poisoned patients, as compared with a control group of average patients. J Orthomol Med 1993; 84: 988-94.

Lindqvist B, Mornstad H: Effects of removing amalgam fillings from patients with diseases affecting the immune system. Med Sci Res 1996; 24 (5): 355-356.

Livardjani F, Ledig M, Kopp P, Dahlet M, Leroy M, Jaeger A: Lung and blood superoxide dismutase activity in mercury vapor exposed rats: effect of N-acetylcysteine treatment. Toxicology 1991; 66 (3): 289-95.

Ngim CH, Devathasan G: Epidemiologic study on the association between body burden mercury level and idiopathic Parkinson's disease. Neuroepidemiology 1989; 8 (3): 128-41

Ngim CH, Foo SC, Boey KW, Jeyaratnam J: Chronic neurobehavioural effects of elemental mercury in dentists. British Journal of Industrial Medicine 1992; 49 (11): 782-90

Oskarsson A, Palminger Hallen I, Sundberg J: Exposure to toxic elements via breast milk. Analyst 1995; 120 (3): 765-70

Oskarsson A, Schultz A, Skerfving S, Hallen IP, Ohlin B, Lagerkvist BJ: Total and inorganic mercury in breast milk in relation to fish consumption and amalgam in lactating women. Arch Environ Health 1996; 51(3) (Arch Environ Health): 234-41

Östlin L. Amalgamutbyte – en väg mot bättre hälsa? En studie om amalgamutbytets hälsoeffekter och försäkringskostnader. Försäkringskassan Stockholm 1991.

Palkiewicz P, Zwiers H, Lorscheider FL: ADP-ribosylation of brain neuronal proteins is altered by in vitro and in vivo exposure to inorganic mercury. Journal of Neurochemistry 1994; 62: 2049-52.

Pendergrass JC, Haley BE: Inhibition of brain tubulin-guanosine 5'-triphosphate interactions by mercury: similarity to observations in Alzheimer's diseased brain. Met Ions Biol Syst 1997; 34: 461-78.

Redhe O: Sjuk av amalgam [Ill from amalgam]. R-Dental AB, Falun, Sweden 1991.

Richardson GM: Assessment of mercury exposure and risks from dental amalgam. Final report. Medical Devices Bureau, Environmental Health Directorate, Health Canada 1995

Richardson GM, Allan M: A Monte Carlo Assessment of Mercury Exposure and Risks from Dental Amalgam. Human and Ecological Risk Assessment 1996; 2: 709-761

Sallsten G, Thoren J, Barregard L, Schutz A, Skarping G: Long-term use of nicotine chewing gum and mercury exposure from dental amalgam fillings. J Dent Res 1996; 75(1): 594-8

Salonen JT, Seppanen K, Nyyssonen K, Korpela H, Kauhanen J, Kantola M, Tuomilehto J, Esterbauer H, Tatzber F, Salonen R: Intake of mercury from fish, lipid peroxidation, and the risk of myocardial infarction and coronary, cardiovascular, and any death in eastern Finnish men. Circulation 1995; 91 (3): 645-55

Seidler A, Hellenbrand W, Robra BP, Vieregge P, Nischan P, Joerg J, Oertel WH, Ulm G, Schneider E: Possible environmental, occupational, and other etiologic factors for Parkinson's disease: a case-control study in Germany. Neurology 1996; 46 (5): 1275-84

Siblerud RL. Health effects after dental amalgam removal. J Orthomol Med 1990; 5:95-106.

Soderstrom S, Fredriksson A, Dencker L, Ebendal T: The effect of mercury vapour on cholinergic neurons in the fetal brain: studies on the expression of nerve growth factor and its low- and high-affinity receptors. Brain Research. Developmental Brain Research 1995; 85 (1): 96-108

Soleo L, Urbano ML, Petrera V, Ambrosi L: Effects of low exposure to inorganic mercury on psychological performance. British Journal of Industrial Medicine 1990; 47 (2): 105-9

Vimy MJ, Takahashi Y, Lorscheider FL: Maternal-fetal distribution of mercury (203Hg) released from dental amalgam fillings. American Journal of Physiology 1990; 258 (4 Pt 2): R939-45.

Vimy MJ, Hooper DE, King WW, Lorscheider FL: Mercury From Maternal "Silver" Tooth Fillings in Sheep and Human Breast Milk: A Source of Neonatal Exposure. Biological Trace Element Res 1997; 56:14352.

Yoshida M, Watanabe C, Satoh H, Kishimoto T, Yamamura Y: Milk transfer and tissue uptake of mercury in suckling offspring after exposure of lactating maternal guinea pigs to inorganic or methylmercury. Archives of Toxicology 1994; 68 (3): 174-8

Warfvinge K, Hua J, Logdberg B: Mercury distribution in cortical areas and fiber systems of the neonatal and maternal adult cerebrum after exposure of pregnant squirrel monkeys to mercury vapor. Environmental Research 1994; 67 (2): 196-208.

Zamm AV: Removal of dental mercury: often an effective treatment for the very sensitive patient . J Orthomol Med 1990; 5 (3): 138-42.

Feedback is welcome!

Last updated: August 6, 1997

Dagfinn Reiersøl, Løvåsveien 3, N-0870 Oslo, Norway
© 1997 Dagfinn Reiersøl