summaries The prognosis of amalgam restorations depends on the alloy composition. Letzel H, Van 'T Hof' M A, Marshall G W, MarshallS}. The influence ofthe amalgam alloy on the survival ofamalgam Restorations: a secondary analysis ofmultiple controlled clinical trials. J Dent Res. 1997; 76: 1787-gB. Objective To assess the influence of alloy composition on the and alloy group. survival of amalgam restorations. Results The total number of failed restorations was 481, of Study selection Data from 14 independent controlled clinical which 77% were restoration-related and 14% process-related. trials on the oral behaviour of Classes I and 2 amalgam Eighty percent of the restoration-related failures were due to restorations, with a follow-up between five and 15 years. some form of fracture of the amalgam. Restorations of conventional zinc-free alloys had the shortest survival. After 13 Data extraction and synthesis All data from the 14 trials were years only 25% survived. Zinc and a high copper content had an cumulated and re-evaluated by secondary analysis. For the equally favourable influence on th€ survival rate, which was 70% analysis, 3119 restorations were available, which were made from after 13 years when either was present. The highest survival rates 24 different alloys by a group of seven operators. The alloys were were of restorations of zinc-containing high-copper alloys: 85% divided into four groups according to their zinc content (zinc­ after 13 years. containing and zinc-free) and their copper content (conventional and high-copper). During the follow-up of the trials, the Conclusions This analysis suggests that the zinc and copper restorations were annually assessed for failures, which were contents of the alloy influences survival of amalgam restorations. classified as to (1) restoration, (2) restorative process and (3) patient-related reasons. With the restoration-related failures, Address for reprints: Dr G W Marshall, Department of Restorative Dentistry, survival functions of the restorations were estimated by alloy Box 0758, University of California, San Fancisco, California 94143-0758, USA

Commentary students and hospital staff and dentists, between survival and operator than to This paper describes a secondary analysis with practically non-existent secondary alloy if exluding the data for the two low­ of fourteen controlled trials carried out caries. The main reasons for restoration copper-no-zinc alloys. However, the between 1974 and 1988-89. A character­ failure were marginal and bulk fractures presentation of the findings only focuss­ istic of these clinical trials is a high scien­ of the material. However, in 'real-life' es on the effect of the alloy, while the tific internal validity due to strict control dentistry the main reason for restoration operator effect is almost ignored. of all potentially confounding factors. replacement is secondary caries2•3 - Although many countries have intro­ There are no indications of selective although it can be pointed out that a duced restrictions on using gamma-2 patient dropout. All patients who entered large part of such replacements may be containing amalgams, these are still the trials received at least six restorations, due to poor differential-diagnosis. 1•2 available. Dentists should carefully con­ which reduced the bias introduced by a The association between the selective sider their rationale for continuing to possible selective dropout. The statistics study population and the failure modes use such alloys in view of the clinical used in the analyses, take into account in the trials may also explain the diverg­ benefit of using high-copper-non-zinc patient dropout by using censored sur­ ing opinions on the association between amalgam alloys. vival time, which reduce the risk of survival and alloy composition.3 A lon­ making wrong inferences from the obser­ gitudinal study reported no association Osborne J W, Summitt J B. The survival of amalgam restorations. J Dent Res. 1998; vation material. There is little reason to between amalgam alloy and survival. 4 77= 340. suspect that the authors' conclusion of an Another study reported equal replace­ 2 Quality evaluations ofdental materials. influence of the amalgam alloy on sur­ ment rates of one conventional and four Anusavice K J (ed). Quintessence vival in this study is just a result of statisti­ high-copper amalgams over 10 years.3 Publishing Co, Chicago, 1989. cal modelling, but supports previous data This study, however, identifies the 3 Jokstad A. Class 2 Cavity Preparations and from Osborne et al. 1 patient caries activity and operator as Restoration Performance. Thesis. University of Oslo: 1992. The low external validity of the trials more important predictors of restora­ 4 Kreulen C Met a/. Replacement risk of due to what the authors characterised as tion failure than the alloy composition. amalgam treatment modalities: 15-year 'a narrowly selective sample' allowed the In Letzel et al, patient caries activity was results.} Dent 1998; 26: 627-32. authors to assess an association between almost absent, which otherwise influ­ Asbjorn Jokstad alloy composition and fractures. The ences restoration survival. In addition, Associate Professor, University of Oslo, patients consisted mainly of dental table 6 displays a stronger relationship Norway

EVIDENCE-BASED DENTISTRY, JUNE 1999 9