Those who are opposed to the use of amalgam often claim that a large number of clinically sound amalgam restorations have recurrent caries underneath them and that resin composite restorations should therefore be preferred. Van Dyke stated, "Almost every amalgam I remove has decay or structural fractures of the tooth."  Dickerson stated, "Over 40 percent of the amalgams deemed in good clinical shape had caries under them."  No studies were cited in the article, but in a personal communication (February 3, 2000), Dickerson cited two studies.
In the first study, Kidd and O'Hara sectioned and histologically examined 30 extracted teeth with marginally defective occlusal amalgam restorations and no clinical evidence of caries. They reported that 54% of the specimens had "microscopic" caries . In the second study, Boston et al microscopically examined two margins for caries on each of 17 teeth extracted for periodontal or prosthetic reasons. The examined teeth showed no clinical evidence of caries. Of the 34 margins, 14 (41%) had microscopic caries .
These studies are of little clinical relevance to the issue of whether amalgam restorations have recurrent caries. The caries in both studies was diagnosable only microscopically and then only after histologic and longitudinal sectioning of the tooth a clinical impossibility. The caries could not be reliably detected with dye staining . If a dentist were to replace an amalgam restoration because it had microscopic caries, it would be clinically impossible to locate the caries during preparation.
There are other reasons not to infer from these laboratory studies of only 17 and 30 extracted teeth that 41% or 54%, respectively, of all amalgam restorations have active caries. The authors of these studies did not make such an assertion. In fact, Kidd and O'Hara noted that "demineralization was slight, no lesion having progressed to cavitation," hypothesizing that the microscopic caries was arrested .
The teeth in both studies were visually inspected for caries, but radiographs were not taken before the microscopic examinations. Dickerson stated that the teeth were "deemed in clinical shape," but most practitioners do not consider teeth with radiographically visible caries to be "clinically acceptable" even if the teeth have no clinically visible caries. The dental histories were not clear. In the first study, the teeth had been extracted for periodontal or prosthetic reasons, but virtually all extracted restored molars are extracted for one of those two reasons. No dental history was reported in the second study. It is therefore unknown how long the amalgams were functioning or what kind of home care the patients had. If most teeth were extracted for periodontal reasons, it is probable that the group of patients whose teeth were used in the study had larger amounts of plaque than do other patients with amalgam restorations.
Based on the year the study by Kidd and O'Hara  was published (1990) and the fact that the amalgam restorations were probably in service for years before extraction, the amalgam probably was not bonded. Amalgam bonding has been shown to decrease microleakage  as much as resin composite bonding . The decrease in microleakage decreases the incidence of caries .
For these reasons, the results of these studies cannot be extrapolated to the general population of patients with amalgam restorations. More importantly, these studies do not prove superiority of nonamalgam alternative restorative materials because there were no non-amalgam (resin composite or cast gold) controls in either study. In a 1989 comparative study of laboratory-induced caries, microscopic caries was significantly more common in teeth with resin composite restorations than in teeth with amalgam restorations . Perhaps microscopic caries after histologic cross-sectioning is common with restorations of all types, and not just in teeth with marginally defective amalgam restorations. In another study in 1990, Kidd et al. showed that much "recurrent" caries is actually residual caries left at the time of preparation, which would have nothing to do with the material chosen to fill the cavity .
Some anti-amalgamists cite a 1994 study  in which the authors removed 16 "clinically sound" amalgam restorations (with no clinical or radiographic sign of caries) in extracted teeth. Dentin in 11 of the 16 was stainable with red caries-detecting dye; in addition, 14 of the 16 (88%) had microscopic evidence of microorganisms.
There are several reasons not to rely on this study as proof that most amalgams have recurrent caries or that resin composite should be preferred over amalgam. First, although each tooth was rinsed with sterile water immediately after extraction and before preparation for the study, it is probable that external microorganisms were still present on the tooth. These organisms could have gotten onto the dentin surface even though sterile burs were used to remove the amalgam and the cavity preparation was cleansed with sterile water. Second, the study was quite small, including only 16 teeth. Third, as with the two previous studies discussed [3,4], there were no resin composite controls to compare for evidence of microorganisms. Finally, and most importantly, the study does not purport to show that most or even any amalgam restorations have recurrent caries underneath them. The authors themselves concluded that caries-indicating dyes do "not discriminate the infection status of the dentin" and that the bacteria in this study "may be due to the effect of long-term microleakage and not due to a residual/recurrent decay process."  In other words, caries was not detected in any of the teeth.
Clinical studies have shown that recurrent caries has been the most common cause of amalgam failure [11-14], but it has also been the most common cause of composite failure [15,12,14-19]. It is even the most common reason for replacement of cast gold restorations . Indeed, secondary caries has been called "an unresolved problem."  However, simply because recurrent caries is the leading reason that restorations need replacement does not mean that most restorations have recurrent caries. Similarly, infections may be a leading cause of failure of dental implants, but most dental implants do not have infections. Most restorative dentists spend about half their time replacing failed restorations , but relatively little time actually examining acceptable restorations . As a result, dentists may perceive that restorations do not last as long as they actually do.
A better way of judging the clinical incidence of recurrent caries is through long-term clinical studies. In a 1989 study of about 600 amalgam-restored teeth recalled at 5 years, none had caries . A 1993 study of 1,415 class II amalgam restorations showed that only 16, less than 2%, had recurrent caries after 10 years . Mair reported that, at 10 years, none of 35 Class II amalgams had recurrent caries. Still another long-term study showed an incidence of less than 5%  after 14 years . Amalgam bonding has been shown to inhibit caries, so the incidence of recurrent caries would probably be even lower today .
A 5-year prospective study of small Class II restorations of primary caries in permanent teeth was published in 1993 by Mjör and Jokstad . They defined small restorations as "having enamel surrounding the cavity margin and with restricted buccolingual extensions of the interproximal and occlusal sections." After 5 years, the composite restorations had a higher incidence of recurrent caries than did the amalgam restorations. Perhaps one reason for this is that the margins of resin composite restorations have been shown to have higher levels of cariogenic bacteria than the margins of amalgam restorations .
Dr. Wahl practices dentistry in Wilmington, Delaware. This article was originally published in Quintessence International 32:525-535, 2001 and is reproduced here with the kind permission of Quintessence Publishing Co. The author also thanks Drs. J. Rodway Mackert, Ivar A. Mjör, and Fred Eichmiller for reading the manuscript and offering several helpful suggestions.