Article Text


Mercury from dental amalgam: looking beyond the average
  1. L Barregard
  1. Correspondence to:
 Professor L Barregard
 Department of Occupational and Environmental Medicine, Sahlgrenska University Hospital and Academy, PB 414, 405 30 Goteborg, Sweden;

Statistics from

Commentary on the paper by Dye et al (see page 368)

Mercury (Hg) is a toxic heavy metal occurring in several physical and chemical forms. Elemental mercury (Hg0) emitted to the atmosphere is converted to soluble forms, deposited into soil and water, and methylated. Methylmercury (MeHg) bioaccumulates up the aquatic food chain and reaches the human diet.1,2 Fish and dental amalgam are two major sources of human exposure to organic and inorganic Hg, respectively.

Even when dental mercury amalgam (at present about 50% mercury by weight) was introduced more than 150 years ago there was concern about its toxicity. After several “amalgam wars”, the safety of dental amalgam was established, however, and no uptake of mercury was supposed to occur from amalgam fillings. The debate was revived more than two decades ago in Europe, and then in North America. In the 1980s several studies confirmed that dental amalgam is a significant source of exposure to mercury in humans, mainly by inhalation of Hg0.1,2 New assessments of exposure and risk were made by national bodies.3,4 Usually the conclusions were that dental amalgam is a source of low level exposure to mercury, but there is no evidence of adverse health effects at these levels. In some countries strong anti-amalgam groups have formed, and policy has changed, aiming at decreasing or abolishing the use of dental amalgam.

The scientific discussion has focused on two questions: What is the dose? Could it cause symptoms and/or disease?


In this issue, Dye and coworkers present recent data on urinary mercury (U-Hg) in a representative sample of about 1600 US women aged 16–49 years from the NHANES study of 1999–2000.5 This is an important piece of information on the exposure to inorganic mercury in the USA. The overall geometric mean (GM) was 0.71 μg/g creatinine (μg/gC) and the arithmetic mean (AM) was 1.1 μg/gC. Since the distribution usually is approximately log normal, the GM is roughly the same as the median. The mean number of amalgam surfaces in posterior teeth was estimated to be 8.7, although the filling material was not registered. (One molar tooth has five surfaces, one of which is occlusal—that is, representing the chewing area.) In women without dental amalgam, the GM was 0.31 μg/gC, and it increased with the estimated number of amalgam surfaces.

These U-Hg levels are similar to those found in Central or Northern Europe; for example, medians of 0.2, 0.3, 0.6, and 1.0 μg/gC in about 4800 Germans with 0, 1–4, 5–8, and >8 teeth filled with amalgam, respectively.6,7 The levels along the Mediterranean coast are usually higher, 1–2 μg/gC in Italy and Portugal, owing to the impact of demethylated MeHg from fish consumption,7 and the same is true for Japan.

For comparison, the TLV (threshold limit value) for occupational exposure to Hg0 is 25 μg/m3 in many countries, which will result in average U-Hg levels of about 30 μg/gC at long term exposure.8 However, in modern industry, the typical U-Hg is lower, and in dentists and dental nurses it is close to the background level in the general population.9,10

Urinary mercury excretion of 1 μg/gC corresponds roughly to an uptake of 3–4 μg inorganic Hg/day, assuming 1.5 g creatinine/24 h and a faecal excretion of mercury similar to that in urine, or higher.

The paper by Dye et al confirms that dental amalgam is a major determinant for U-Hg. However, the bottom line (“main message”) that 10 posterior amalgam surfaces would result in an increase in U-Hg of 1.8 μg/gC must be questioned. This figure was obtained simply by multiplying the regression coefficient in table 5 (0.06) by 10, and taking the antilog of 0.6, which is 1.8.

But if lnU-Hg increases with 0.6, then the untransformed U-Hg will be multiplied by 1.8, and the increase with 10 posterior surfaces will depend on the starting point. The intercept is important at low levels and the model chosen assumes a curvilinear (exponential) increase. It is not possible to state in general terms how much U-Hg will increase with 10 additional amalgam surfaces. When the background concentration (given by the authors for dentate women with no dental restorations) is 0.31 μg/gC, 10 additional surfaces will result in an increase of the geometric mean by 0.25 μg/gC. Similarly, if the GM U-Hg is 0.71 μg/gC, as in the “average US woman” with nine surfaces, and 10 more surfaces are added, the model will predict an increase of 0.57 μg/gC. The corresponding AM will be somewhat higher, but much lower than 1.8 μg/gC. In fact, the study in US men referred to by the authors11 found an increase of 0.6 μg/gC per 10 surfaces, which is similar to my interpretation of the NHANES data.

However, the regression coefficient for a continuous variable can be transformed into a percent increase per surface. The study by Dye et al estimates an increase by about 6% per posterior amalgam surface. This is higher than previously reported.11,12


Mild subclinical effects on the CNS may occur at occupational exposure levels with U-Hg of about 25 μg/gC.1 Another early effect commonly reported is that on the renal proximal tubules. Small molecular weight proteins and lysosomal enzymes are found in urine in increased amounts. Such effects have been reported at occupational exposure at U-Hg as low as 10–20 μg/gC.13 Although some studies have suggested effects on the CNS at even lower occupational exposure in dentists,9 this has not been confirmed by others,10 and the debate about dentists is still ongoing.2,14 As Dye et al point out, long term follow up in children treated with dental amalgam or composites (“controls”) is being performed in two controlled trials. These studies, the Children Amalgam Trial (CAT study) in Boston and Maine, and the Casa Pia study in Portugal, will hopefully shed more light on this issue with respect to low level exposure early in life.

Even if we consider the evidence of effects below 10 μg/gC too weak, does it imply that mercury from dental amalgam is always safe? Not necessarily, if we look beyond the average. Taking the overall SD for lnU-Hg given by Dye et al (0.96), an ideal log normal distribution predicts the 99 percentile to be 6.6 μg/gC and the 99.9 percentile 14 μg/gC. This fits well with the right tail actually found (6 μg/gC and 14 μg/gC even if three higher outliers were excluded (B Dye, personal communication). It implies that about one in 1000 US women of this age have a urinary Hg higher than in many settings with occupational mercury exposure, and similar to levels that result in discrete effects on the kidneys. In the study of US men the 99 percentile was about 7 μg/gC and the 99.9 percentile about 20 μg/gC.11

Similar estimates for the Swedish general population were presented in this journal in the 1990s.15 Owing to a higher amalgam load (typically 30 surfaces) and higher U-Hg levels (geometric mean 2 μg/gC) in Swedish subjects, the 99 percentiles and 99.9 percentiles were then estimated at 10 and 25 μg/gC, respectively.

The variability of U-Hg is not caused only by the variability in number of amalgam surfaces. Cases with high U-Hg levels may be found among heavy chewing gum users; for example, a median of 11 μg/gC (maximum 25 μg/gC) in 18 long term users of nicotine chewing gum.16 Urinary mercury as high as 50 μg/gC may be caused by dental amalgam.15 High levels were also reported in people eating large amounts of contaminated fish; for example, a median of 7 μg/gC (maximum 22 μg/gC) in 22 men with high consumption of tuna,17 or in those who use skin lightening creams or other mercury containing ointments.

If we accept that dental amalgam in rare cases, for example, at intense long term chewing, may result in high exposure with U-Hg levels of 10–50 μg/gC, we should accept that in those cases it could slightly affect the kidneys and the CNS.

In summary, the average mercury dose from amalgam fillings is in most cases below 5 μg per day on average,1 but the issue of the right tail of the distribution is more important. A substantial number of persons have a mercury uptake from dental amalgam that is higher than should be accepted. Many of these cases are likely to be found among people with intense chewing gum use. Fortunately an increased Hg uptake can easily be shown in a urine sample.


I thank Dag Ellingsen and Gerd Sällsten for their comments on the draft manuscript.

Commentary on the paper by Dye et al (see page 368)


View Abstract


  • Competing interests: none

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles