While several studies have been published on the longevity of amalgam restorations and crowns before failure, most are difficult to compare because they vary in the number of patients, years of follow-up, definition of failure, number and type of clinicians, number of restorations per patient, type and size of restorations and type of statistical methods used.^17–30 Owing to variation in studies and lack of parameters,³¹ there is no consensus in the literature on the longevity of large amalgam restorations and crowns and the criteria for placing one versus the other.

Three studies have evaluated the outcomes of large amalgam restorations and crowns.^28–30 Hawthorne and Smales²⁸ reviewed the dental records of 100 patients in three private practice offices in Adelaide, Australia. They considered a restoration to be a failure if it was replaced fully or partially owing to caries or fracture or if the tooth was extracted owing to caries. The estimated median survival time was 22.5 years for amalgam restorations and 26 years for crowns. The reported survival time for amalgam restorations included amalgam restorations of all sizes and, therefore, is hard to compare with studies of large or cuspal covered amalgam restoration.

The two study questions were what factors were related to whether a tooth received any subsequent treatment during a 10-year period and what factors were related to whether a tooth received catastrophic treatment during a 10-year period.

In another study, Smales and Hawthorne²⁹ found that crowns were significantly more likely to survive over time (that is, no marginal caries, repair, replacement, or loss from fracture or cementation failures) than were large amalgam restorations that replaced one or two cusps. They estimated that 96 percent of cast gold crowns survived at five years, 91 percent at 10 years and 78 percent at 15 years. In contrast, they estimated that 78 percent of amalgam restorations survived at five years, 67 percent at 10 years and 48 percent at 15 years.

Martin and Bader³⁰ followed four- and five-surface amalgam restorations for five years and found similar, yet less drastic, differences between crowns and large amalgam restorations using a slightly different definition of success (needing no further treatment or an additional one- or two-surface restorations). After five years, 84 percent of porcelain and gold crowns were successful, with 8 percent of the porcelain crowns and 11 percent of the gold crowns having a catastrophic event (that is, tooth needing endodontic treatment or extraction). In contrast, 72 percent of the four-surface amalgam restorations and 65 percent of the five-surface amalgam restorations were successful, with 10 percent of the four-surface and 15 percent of the five-surface amalgam restorations experiencing a catastrophic failure. Neither Smales and Hawthorne²⁹ nor Martin and Bader,³⁰ however, evaluated the influence of patients’ oral and physical health factors on the longevity of the restorations.

To further understand the factors that affect the longevity of large amalgam restorations and crowns, we compared the subsequent treatment given to teeth that had large amalgam restorations with that given to teeth that had received both a large amalgam restoration and a crown over a 10-year period. The two primary research questions in our study were what factors were related to whether a tooth received any subsequent treatment during a 10-year period and what factors were related to whether a tooth received catastrophic treatment during a 10-year period. Our primary research hypothesis was that teeth that had received a crown would have less subsequent treatment and less catastrophic treatment than teeth treated with a large amalgam restoration alone. We defined tooth outcomes as "no subsequent treatment," "minor subsequent treatment" and "catastrophic subsequent treatment." We then evaluated factors associated with the subsequent treatment received by teeth with the two restorations.

	MATERIAL AND METHODS

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The Committee for the Protection of Human Subjects at the University of Iowa approved our study. We used retrospective data from both the administrative database of the University of Iowa College of Dentistry (UICD) and patient records to evaluate the long-term outcomes of large amalgam restorations and crowns for 518 routine users. We defined routine users as patients who had at least one visit every two years. We selected teeth that received restorations that covered four or more surfaces in 1987 or 1988 as "target" teeth and evaluated them over a 10-year period. We categorized each tooth as either a crown (if the tooth received a crown within one year from the placement of the large amalgam restoration) or a large amalgam restoration (if the tooth did not receive a subsequent crown in the following year).

Data collection and definition of variables. We extracted data for treatment provided between mid-1985 and mid-2001 for all teeth that received a large amalgam restoration at UICD in 1987 or 1988. To avoid the use of related data, we randomly selected only one tooth per patient and included only restorations on permanent molars and premolars. We excluded teeth that had received endodontic therapy before the placement of the large amalgam restoration on the target tooth, as well as teeth that served as a bridge abutment after the placement of the large amalgam restoration or crown.

Teeth with large amalgam restorations were more likely to have had caries before the placement of the initial large amalgam restoration than were teeth with crowns.

We categorized the outcome for each tooth as having received no treatment, minor treatment (restorative treatment not including endodontic therapy) or catastrophic treatment (endodontic therapy or extraction). To address the first research question, we combined teeth that had received either minor or catastrophic treatment into a category called "any treatment" and compared them with teeth that had received no treatment. To address the second research question, we combined teeth that had received either no treatment or minor treatment and compared them with teeth that had received catastrophic treatment.

Table 1 lists the characteristics we considered as possible explanatory variables related to the subsequent treatment received, if any, by each tooth. The type of restoration (large amalgam restoration or crown) was the primary explanatory variable.

We used multiple logistic regression to find the best predictive models for the receipt of any treatment and the receipt of catastrophic treatment. We used results from bivariate analyses as inclusion criteria for multiple logistic regression models. We included variables with a P value .2 in the initial logistic regression models. We then conducted both a forward stepwise approach and backward elimination approach. We compared the results of the forward stepwise approach and backward elimination models and explored discrepancies. We used P = .1 as the significance level for a variable to be kept in the final model. We evaluated age as both a continuous variable and in a quadratic form for inclusion into the models. We also investigated interactions between the remaining explanatory variables and restoration type.

	RESULTS

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Table 1 shows the characteristics of patients and teeth that initially had received large amalgam restorations or crowns. The 518 target teeth were distributed evenly between large amalgam restorations (49.2 percent) and crowns (50.8 percent). We found some differences in the characteristics of the patients. People with crowns were more interested in keeping their teeth, had more teeth present in their mouths and had fewer visits per year on average but more recall visits on average than did people without crowns. The teeth with crowns were more likely to be premolars than molars, have pins and a greater number of pins, have two approximal teeth and have prior restorations than were teeth with only a large amalgam restoration. Teeth with large amalgam restorations were more likely to have had caries before the placement of the initial large amalgam restoration. The number of positive responses to the presence of sensitivity and periodontal pockets was low and, therefore, we did not include them in the multiple logistic regression analysis.

Receipt of subsequent treatment. Table 2 (page 743) displays the distribution of each explanatory variable for teeth that had received no treatment (n = 297), minor treatment (n = 134) or catastrophic treatment (n = 87). About 43 percent of all teeth received some treatment during the 10-year period. This varied significantly by restoration type, with 64 percent of teeth with large amalgam restorations receiving some subsequent treatment, compared with 32.1 percent of teeth with crowns. While most of the subsequent restorative procedures in the teeth with large amalgam restorations were minor (42 percent), twice as many teeth with large amalgam restorations received catastrophic treatment as did teeth with crowns.

Initial logistic regression models. Table 3 shows the resulting odds ratios, 95 percent confidence intervals and P values for each variable in the initial multivariable models for whether the teeth received any subsequent treatment or catastrophic treatment. Teeth with large amalgam restorations; men; patients with a history of grinding or clenching, having fewer teeth present or having a higher average number of visits per year; mandibular teeth; having zero teeth or one approximal tooth; not having an opposing tooth; having caries when the initial large amalgam restoration was placed; and not having a broken tooth or restoration all were associated with having any subsequent treatment.

Final logistic regression model—any subsequent treatment. Table 4 presents significant predictor variables from the final regression model for teeth that received any subsequent treatment. While controlling for age, sex, grinding or clenching, and broken tooth or restoration, we found that the odds of a tooth with a large amalgam restoration having any treatment were 6.8 times the odds of a tooth with a crown receiving any treatment over the 10-year period. Similarly, the odds of people with a history of grinding or clenching having any treatment were 1.8 times the odds of people without a history of grinding or clenching having any treatment. Lastly, the odds of a broken tooth or restoration having any treatment were 1.4 times less than the odds of a tooth that was not broken having any treatment.

	DISCUSSION

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
With the objective of following the long-term treatment histories of teeth restored with large amalgam restorations and crowns, our study is different from most other published studies on restoration longevity that follow teeth only to the first subsequent treatment. Our definition of failure or success also provides better clinical relevance regarding whether a restored tooth received any treatment or catastrophic treatment that can be useful for discussions with patients.

Teeth with large amalgam restorations and crowns received less subsequent treatment of any kind than did teeth with large amalgam restorations only. This could be owing to both the longevity of the crown and appropriate treatment planning given the higher cost of the crown.

Crowns were more likely to be placed initially in people with characteristics associated with regular dental users—people who were more interested in keeping their teeth, had more teeth present, had fewer visits per year on average but had more recall visits on average and were less likely to have caries. Thus, the teeth with crowns were less likely to require any treatment and less catastrophic treatment over 10 years. These results are similar to those of a comparable study by Martin and Bader,³⁰ who found less subsequent treatment for teeth with crowns.

Patients who had a history of grinding were more likely to have had subsequent treatment, yet grinding was not related to the receipt of catastrophic treatment. The finding of a relationship of grinding to restoration outcome is supported by Burke and colleagues.³² They found that normal occlusal function was associated with increased age of restoration at replacement, and excessive and high occlusal function was associated with reduced age of restoration at failure.

Having a broken tooth or restoration, rather than just caries, at the time of the placement of the large amalgam restoration appeared to act in a protective manner, with broken teeth having less subsequent treatment than did teeth that were not broken at the time the large amalgam restoration was placed. The type of restoration was the only patient factor significantly associated with having catastrophic subsequent treatment.

A major conceptual limitation in our study was the inability to factor in the effect of the difference in the cost between the two procedures on evaluating the difference in outcomes. As a crown can cost five times as much as a large amalgam restoration to place originally, cost must be considered during the treatment planning process. We are conducting a cost-effectiveness analysis using similar data to factor costs into the decision-making process.

Our study included care provided by a large number of faculty, graduate students and dental students in a variety of clinics. While there may be general treatment philosophies on when to place a crown versus a large amalgam restoration, it is expected that differences exist regarding treatment planning throughout dental schools over time, as has been found in private practice settings. Having diagnosis codes for why dental procedures are completed would help with all dental outcome studies using administrative data to determine factors related to the longevity of a restoration.

The receipt of subsequent treatment should not necessarily be interpreted as "need" for treatment at the individual level. "Need" implies a gold standard that is not available at the individual level at this time. Further research on outcomes of commonly provided alternative treatment modalities and the importance of diagnostic coding for evaluating these outcomes can help with this issue.

The findings of our study suggest that it is appropriate to consider patient and oral health characteristics in the clinical decision-making process when faced with a tooth that has lost a significant amount of structure. Additionally, our study provides information that can be used to facilitate patient education of what factors to consider when choosing between a large amalgam restoration and a crown.

Since we investigated the receipt of subsequent treatment in a group of patients who continuously sought treatment in a dental school for at least 10 years, the results of our study should be generalized beyond this population carefully. Treatment planning in a dental school may be different from that in a private practice setting, and the patient population in a dental school may be fundamentally different from that in a private practice. To be confident that we obtained all of the procedures provided to the target teeth, we collected data only from patients who sought care continuously. By using data only from continuous users of care, we created a biased group of patients. Information on patients who did not seek continuous care was not available; therefore, we do not know how these patients are similar to or different from patients who seek continuous care.

We may have created bias in determining if the restoration type was a large amalgam restoration or a crown. We did not have information on the original intent of the large amalgam restoration—whether it was to serve as a buildup for a crown or as the definitive restoration. Therefore, as a proxy of the original intent to place a crown, we classified teeth as a crown if a crown was placed within one year of the large amalgam restoration. If the treatment plan for a tooth with a large amalgam restoration originally did not include placement of a crown, but ended up including crown placement within a year anyway, our data would be misclassified in this respect and would underestimate subsequent treatment received by the tooth. Conversely, if the crown was planned but delayed beyond a year, the amount of treatment for the large amalgam restoration would have been overestimated. It also should be understood that the administrative and patient record information originally was not intended to be included in a research study such as this one; therefore, there may have had some inconsistencies in coding and documentation.

Several pieces of desirable information about considerations in the treatment planning process were not consistently available in either the administrative data or the patient record. This information included the reason for the selection of the large amalgam restoration or crown (for example, was it the dentist’s preference, the patient’s preference or both), the level of insurance coverage for crowns, personal income information, more information about the type and size of amalgam restoration that was placed, and more information about the oral health status of the patient. To further address the long-term outcomes of large amalgam restorations and crowns, we have evaluated the treatment patterns over time,³³ the timing of subsequent treatment³⁴ and the costs associated with treatment selection.

	CONCLUSIONS

TOP ABSTRACT MATERIAL AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
After the placement of a large amalgam restoration, teeth that received a crown within one year were less likely to receive any subsequent treatment and were less likely to receive catastrophic treatment over a 10-year period than were teeth that received only a large amalgam restoration. Both patient and tooth factors were related as to whether a tooth received any further treatment, while only restoration type (large amalgam restoration versus crown) was related to whether a tooth received a catastrophic treatment. The difference in the long-term outcomes of these types of restorations may be related to the way treatment was planned for teeth with poorer prognoses for large amalgam restorations rather than crowns.