All C opyrig

eR

ht
ech
te

by
vo
rbe

Qu
ha
nt lte
n
Longevity and Reasons for Failure of Sandwich and

i
e ss e n z
Total-etch Posterior Composite Resin Restorations
Niek J. M. Opdama/Ewald M. Bronkhorsta/Joost M. Roetersb/Bas A. C. Loomansa

Purpose: To investigate longevity and reasons for failure of Class II posterior composite restorations (PCRs) placed
with or without a lining of glass-ionomer cement.

Materials and Methods: Four hundred fifty-eight Class II PCR placed in 248 patients (110 male, 138 female, age 18
to 80) by two dentists in a general practice between 1988 and 1997 were retrospectively examined from the patient
files. The restorations were placed either with a total-etch technique or with a resin-modified glass-ionomer lining
placed on the dentin. Items recorded were date of placement, date of last check-up visit, tooth number, and restored
surfaces. Date of replacement and reason for failure of the PCR was recorded. A restoration was clinically acceptable
when still in function and acceptable at the last check-up visit. Additionally, the caries risk for each patient was esti-
mated by the treating clinician. Life tables and Kaplan-Meier curves were used to express survival rates. A Cox regres-
sion was applied to assess the influence of variables on survival.

Results: Three hundred seventy-six total-etch PCRs and 82 PCRs with a lining were investigated. After 9 years, sur-
vival percentages of 88.1% for total-etch restorations and 70.5% for restorations with a resin-modified glass-ionomer
lining were found. The most important reasons for failure were fracture and caries. Predominantly, failures started oc-
curring after 3 to 4 years of clinical service. Results of the Cox regression show that the presence of a lining and high
risk for caries significantly increased the failure rate of the restorations.

Conclusion: PCRs placed with a resin-modified glass-ionomer lining clinically showed more frequent fractures than
PCRs placed with a total-etch technique.

Keywords: composite, glass ionomer, lining, longevity, clinical, Class II, restrospective, failure, sandwich.

J Adhes Dent 2007; 9: 469-475. Submitted for publication: 14.07.06; accepted for publication: 11.07.07.

C omposite resin has become an accepted restorative ma-

terial for Class II lesions in premolars and molars.
Prospective clinical studies on posterior composite resin
standard procedure to obtain bonding of the composite resin
to dentin. Two variations of that type of restorations exist: the
open and closed sandwich. In a closed sandwich, the dentin
restorations show an annual failure rate of 1% to 4% de- is covered with a resin-modified glass-ionomer (RMGI) lining
pending on the type of study and selected materials.16 The cement. In an open sandwich, a glass-ionomer restoration
improved reliability of composite resin as a restorative ma- cement is used to replace the dentin and also to fill the cer-
terial in posterior teeth largely depends on the combination vical part of the box, which results in a substantial part of the
with an adhesive technique, which results in reduced mi- glass-ionomer cement being exposed to the oral environ-
croleakage and strengthening of weakened tooth structures. ment. Longevity studies of posterior composite resin restora-
In the years when reliable dentin adhesive systems were not tions placed with a closed sandwich technique generally
available, application of a glass-ionomer cement was the demonstrate a good survival rate.8,15 Long-term data on
open sandwich restorations are few, but show an annual fail-
ure rate of 1% to 3%.2,12,25
a Assistant Professor, Department of Preventive and Curative Dentistry, Rad- Since the introduction of reliable dentin adhesives, the
boud University Nijmegen Medical Center, Nijmegen, The Netherlands.
application of a liner or base on the dentin for protection of
b Associate Professor, Department of Preventive and Curative Dentistry, Rad- the pulp or to establish adhesion is no longer considered
boud University Nijmegen Medical Center, Nijmegen, The Netherlands.
necessary. Generally, posterior composite resin restorations
Reprint requests: Niek JM Opdam, Department of Preventive and Curative
Dentistry, Radboud University Nijmegen Medical Center, P.O. Box 9101, NL
placed without a liner show good longevity.6,13,21,26 Howev-
6500 HB Nijmegen, The Netherlands. Tel: +31-24-361-6410, Fax: +31-24-354- er, the sandwich technique is nowadays still recommended
0265. e-mail : n.opdam@dent.umcn.nl by some authors, as it is thought to have some advantages.

Vol 9, No 5, 2007 469

 All C opyrig
eR

ht
ech
te

by
vo
rbe

Qu
Opdam et al
ha
n lte
n
te

i
A lining with a low modulus of elasticity, such as a glass- I and II composite resin restorations placed between 1988
ionomer cement, is expected to act as a stress-absorbing lay- and 2002.19 ss e n z
er and to compensate for polymerization shrinkage stress.9 From this database, all Class II composite resin restora-
This elastic layer concept was introduced in the early 90s, tions placed between 1988 and 1997 according to one of
and several in vitro studies have shown that the application the two following protocols were selected:
of a lining might be advantageous to reduce microleakage
around Class II composite resin restorations,5,11,18 while oth- 1. Restorations in which a closed-sandwich technique was
er studies failed to demonstrate a positive effect.1,7 Anoth- applied. A layer of RMGI lining cement – either Vitrebond
er advantage of glass-ionomer cement might be the pres- (3M ESPE; St Paul, MN, USA) or GC lining (GC; Tokyo,
ence of fluoride, which may contribute to a reduction of sec- Japan) – was placed in a thin layer on the entire dentin
ondary caries. However, a systematic review on clinical stud- surface followed by etching the cavity walls with 38%
ies with glass-ionomer cement restorations failed to find ev- phosphoric acid for 15 s, rinsing, and drying. Then the ad-
idence of this effect,22 and no clinical studies on the pres- hesive resin (Clearfil Photo Bond, Kuraray; Osaka, Japan)
ence of secondary caries in relation to posterior composite was applied, gently air dried, and cured for 10 s. The com-
resin restorations with and without a glass-ionomer lining posite resin (Clearfil Photo Posterior, Kuraray) was inject-
are available. ed into the cavity and placed and cured in increments.
A possible disadvantage of the application of a lining ce- 2. Restoration in which a total-etch technique was applied.
ment with its low modulus of elasticity is a weakening effect 38% phosphoric acid was placed simultaneously on
on the strength of the overlying composite resin. In vitro enamel and dentin for 15 s. Then the cavity was rinsed
studies have shown that posterior composite restorations and air dried before application of the dentin primer
placed without a liner are significantly stronger than those (Clearfil SA primer, Kuraray), which was carefully dried.
placed with a liner of glass-ionomer cement.3 Another in vit- Then the adhesive resin (Clearfil Photo Bond, Kuraray)
ro study demonstrated that a layer with a low E-modulus ma- was applied, gently air dried, and cured for 10 s. The
terial placed between two high E-modulus materials (tooth same composite resin (Clearfil Photo Posterior, Kuraray)
and composite resin) leads to a concentration of forces in was injected into the cavity and placed and cured in in-
the elastic layer when the tooth is loaded.24 Furthermore, in crements.
vitro studies showed that in restorations with a lining of a low
E-modulus material, the microleakage increased after sub- Four hundred fifty-eight Class II restorations were placed
jecting the restorations to loading.4,10 In a clinical situation, according to these protocols in 248 patients (110 male, 138
this could result in restorations placed with a lining being female, age between 18 and 80 years). The Class II restora-
more sensitive to fatigue after repeated loading. This could tions varied from rather small box-type Class II restorations
lead to higher clinical failure of the restorations either due in primary proximal caries to large cusp-replacing restora-
to fracture or caries. tions. The decision to use a glass-ionomer cement as lining
To mention another confounding factor, the development did not depend on the size of the preparation, but rather on
of secondary caries also depends on whether the patient is the period in which restorations were placed, as in the years
a caries-risk or non-risk patient. This probably influences the 1990 to 1995, it gradually became “state of the art” to use
survival and type of failure of sandwich or total-etch restora- the total-etch technique instead of the sandwich technique.
tions, as the presence of glass ionomer might contribute to No calcium hydroxide lining was applied in the cavities. The
lessening failures due to secondary caries. A simple method restorations were placed by one of two operators. The first
has been described to classify patients according to the ex- operator was an experienced dentist who graduated in
pected caries risk,17,23 which has been used in another clin- 1980. He placed 360 restorations from 1988 to 1997. The
ical study.13 second operator started in the practice in 1992 immediate-
Until now, no long-term data have been available which ly after graduation, and placed 98 restorations from 1992
compare longevity and reasons for failure for composite to 1997.
resin restorations placed with a sandwich or total-etch tech- From the files, the date of placement of the restorations
nique. The aim of this retrospective clinical study was to in- and the date of the last check-up visit were recorded up to
vestigate differences in longevity and the reasons for failure December 2005. Other recorded items were the operator in-
of Class II posterior composite restorations placed with or volved, the tooth number, and the restored tooth surfaces.
without a lining of glass-ionomer cement in patients with a If a restoration was replaced, the date of replacement and
low or high caries risk. reason for failure of the restoration were recorded. When a
restoration was still in function at the date of the last check-
up visit in the practice, the restoration was considered to be
MATERIALS AND METHODS clinically acceptable. The examination during the last check-
up visit was done by one operator (operator 1).
In a previous longitudinal study, amalgam and composite Teeth that were restored as a pretreatment to serve as an
restorations placed in a general practice by two operators abutment tooth for crowns were excluded from the study.
were retrospectively evaluated from the patient files. In- Whenever a tooth received a new restoration without affect-
cluded in the study were files from all patients who regular- ing the existing restoration, eg, when caries on a distal sur-
ly attended the dental practice for check-ups and follow-up face of a tooth with a mesial box restoration was restored
treatment in 2002. This resulted in a large database of Class with a distal box-type restoration, this was not considered as

470 The Journal of Adhesive Dentistry

 All C opyrig
eR

ht
ech
te

by
vo
rbe

Qu
Opdam et al
ha
nt lte
n

i
Table 1a Size of the restorations according to number of treated surfaces e ss e n z
2-surface 3-surface 4/5-surface Total
Total-etch restorations 138 166 72 376
Sandwich restorations 40 33 9 82
Total 178 199 81 458

Table 1b Size of the restorations according to number of treated surfaces

Sandwich Total-etch Total

Operator 1 39 322 361

Operator 2 43 54 97
Total 82 376 458

a failure. However, endodontic treatment, replacement due “operator” was added as a co-variable to the multivariate
to pain and extraction, although not necessarily related to model. By doing so, the estimates of the explicatory vari-
the restorations, were always considered failures. In the ables is corrected for potential operator effects. To eliminate
same way, a cusp fracture next to an existing restoration, al- variables with no significant effect on restoration survival, a
though often repaired without a total replacement of the backward regression procedure was applied. Cox regression
restoration, was considered to be related to the existing was applied to assess the influence of these variables on
restoration and considered a failure. restoration survival. Since for some patients more than one
Additionally, the caries risk for each patient was estimat- restoration was included in the study, the condition stating
ed retrospectively by the treating clinician (the first operator) that all measurements are independent was violated. There-
by means of clinical and sociodemographic information rou- fore, the results of the Cox model were checked with an ex-
tinely available at the annual clinical examinations, eg, in- tended Cox model, containing a gamma frailty term. This last
cipient caries lesions and former caries history.12 model is implemented using the statistical software R (Lan-
The data set was statistically analyzed using SPSS soft- guage and Environment for Statistical Computing, Founda-
ware. The analysis consisted of two parts. The first part had tion for Statistical Computing; Vienna, Austria).
a descriptive character and was used to describe the mate- Because it was possibly a confounding factor that in the
rial and allow for easy comparison with existing literature on period in which both sandwich and total-etch restorations
this subject. For this purpose, cross tabs were made to il- were placed, the selection made by the dentist could have
lustrate the various types of restorations made; life tables played a role (eg, difficult cases preferably sandwich, easy
were used to calculate survival rates and annual failure cases total-etch), a Cox regression was also used to compare
rates (AFR) according to the formula data for total-etch restorations placed in the period 1994 to
1995 with those placed in 1996, when the sandwich tech-
√ 1 - AFR
k
P= nique was no longer used.

in which p expresses the percentage of restorations surviv-

ing after k years of service. Kaplan-Meier analysis was used
to create survival curves. RESULTS
The second part was a statistical analysis aimed at quan-
tifying the effect of lining cement on the survival of Class II The survival rates of 376 total-etch Class II restorations and
composite resin restorations. Since this survival is likely to 82 sandwich restorations were investigated. In Table 1a, the
be influenced by more than just the presence of lining ce- distribution of two-, three- and more surface restorations
ment, a multivariate Cox regression was applied. The fol- among groups are presented. In Table 1b, the restorations
lowing variables were available to explain the survival of the placed by the two operators are shown. The observation
restorations made: lining cement (yes, no), number of sur- time was 8.93 (mean) and 9.46 (median) years for the sand-
faces, operator 1 vs 2, premolar vs molar, caries risk of the wich restorations and 8.57 (mean) and 8.72 (median) years
patient (low/high), patient age, patient gender, and year of for the total etch restorations. During the observation peri-
treatment. The inclusion of the variable “operator” is note- od, 43 total-etch restorations and 34 sandwich restorations
worthy. Since the number and type of restorations made was failed. In Table 2, the main reasons for failure and the dis-
not balanced with regard to operators, the variable “opera- tribution of restorations among caries-risk and non-risk pa-
tor” might act as a confounder in this study. To avoid this, tients are shown.

Vol 9, No 5, 2007 471

 All C opyrig
eR

ht
ech
te

by
vo
rbe

Qu
Opdam et al
ha
n lte
n
t eofs

i
Table 2 Number of clinically acceptable and failed restorations and reasons for failure, related to caries risk
patients se nz
Total-etch restorations high caries risk low caries risk total nr. total-etch
% n n % n n % n
clinically acceptable 78% 102 94% 231 89% 333
failed – caries 16% 21 2% 5 7% 26
failed – fracture 1% 2 2% 4 1% 6
fracture of restoration 1 3
fracture of tooth 1 1 3% 11
Other reasons for failure 2% 5 2% 6
pain/endo 1 4
extraction 4 1
loss of retention 0 0
unknown 0 1
Total 130 246 376

Closed sandwich restoration high caries risk low caries risk total nr. sandwich
% n n % n n % n
clinically acceptable 65% 20 55% 28 59% 48
failed due to caries 26% 8 6% 3 13% 11
failed due to fracture 10% 3 29% 15 22% 18
fracture of restoration 2 5
fracture of tooth 1 10
Other reasons for failure 0% 0 10% 5 6% 5
pain/endo 0 2
extraction 0 1
loss of retention 0 1
unknown 0 1
Total
31 51 82

Table 3 Survival and annual failure rate

Total-etch restorations Closed-sandwich restorations

Caries risk of patients high risk low risk high+low risk high risk low risk high+low risk
Annual failure rate 2.7% 0.8% 1.4% 2.8% 4.4% 3.8%
% in service after 9 years 78% 93.3% 88.1% 77.0% 66.5% 70.5%

The survival rates and annual failure rates after 9 years results for the final Cox regression are shown in Table 4. All
are listed in Table 3. After 9 years, a survival percentage of included variables showed statistical significance. Espe-
88.1% for the total-etch restorations was found, compared cially a sandwich restoration placed in a high caries-risk pa-
to 70.5% for the sandwich restorations. The Kaplan-Meier tient significantly increased the failure rate of the restora-
survival curves for both groups of restorations are shown in tions. The odds ratio of 4.08 indicates that, while correcting
Fig 1. for the other variables in the regression model, sandwich
Figures 2a and 2b present the relation between the time restorations have a failure risk about four times as high as
and type of failure. The most important reasons for failure total-etch restorations. Smaller differences were found be-
are: fracture of the tooth and/or restoration and caries. Pre- tween operators, number of treated surfaces, and between
dominantly, these failures started occurring after 3 to 4 molar and premolar teeth.
years of clinical service. The total-etch restorations placed in 1996 did not per-
The results of the extended Cox model including a gam- form significantly differently from those placed in the period
ma frailty term were identical to that of the basic Cox mod- 1994 to 1995 (Cox regression corrected for number of ser-
el. This implied that the clustering of some of the restora- vices: Exp (B) = 1.264, p = 38.3). Thus, selection made by
tions did not influence the results of the Cox regression. The the dentist was excluded as a possible confounding factor.

472 The Journal of Adhesive Dentistry

 All C opyrig
eR

ht
ech
te

by
vo
rbe

Qu
Opdam et al
ha
nt lte
n

i
e ss e n z

Fig 1 Kaplan-Meier survival curves for total-etch restorations and

restorations placed with a glass-ionomer lining (sandwich restora-
tions).

years years

Fig 2a Type of failure and functional time of each failed total- Fig 2b Type of failure and functional time of each failed sand-
etch restoration (43 failed of 376 placed restorations). wich restoration (34 failed of 82 placed restorations).

Table 4 Final model of backward Cox regression analysis

Variable OR 95% CI Significance

Lining cement (1=yes; 0=no) 4.08 2.43-6.85 < 0.001

Operator 0.51 0.27-0.99 0.047
Number of surfaces 1.39 1.01-1.93 0.045
Premolar (1) vs molar (0) 0.46 0.27-0.79 0.005
High caries risk (1=yes; 0=no) 2.45 1.55-3.88 < 0.001

Vol 9, No 5, 2007 473

 All C opyrig
eR

ht
ech
te

by
vo
rbe

Qu
Opdam et al
ha
n lte
n
te

i
DISCUSSION It is often suggested that glass-ionomer cement should
lead to a prevention of secondary caries. Moreover, nz
ssa elining
In the present study, the longevity of posterior composite of RMGIC is considered to act as an elastic layer, which
restorations placed with either the total-etch technique or a should improve adaptation of the restoration and compen-
sandwich technique with a resin-modified glass-ionomer lin- sate for polymerization stress. The results of this study do
ing was compared. The retrospective examination of well- not support either assumption. The occurrence of sec-
kept and detailed patient files enabled the evaluation of a ondary caries appeared to depend on the caries risk level
large number of restorations placed in a general dental prac- of the patient and not on the presence or absence of glass-
tice. Randomized clinical trials have the advantage of stan- ionomer cement. This finding is in accordance with a sys-
dardization of methods and calibration of the operators, al- tematic review that failed to support the statement that
lowing a more reliable comparison between different treat- glass-ionomer cement has a cariostatic effect under clini-
ment options. However, such a study design does not reflect cal conditions.22
the situation in a general dental practice, as patients are se- The results of the present study support the assumption
lected and operators are aware of the fact that restorations made in the introduction that the presence of a lining ce-
will be evaluated. This might explain why the results of ran- ment reduces the fracture resistance of a restoration, as a
domized clinical studies are often better than those of ret- significantly higher failure rate was found for these type of
rospective studies. In the present study, a variety of patients restorations in the present study. The increase in failure of
was included, thus, high-caries risk patients were also in- the sandwich group was mainly due to fracture either of the
cluded in the study population. According to the general restoration or of the tooth, or both. Clinical findings in oth-
dentist who treated the patients in the period of the study, er studies on sandwich restorations are in accordance with
161 restorations were placed in high-risk patients while the the present findings. In one study, the 6-year results show
other 297 were placed in low-risk patients. an increasing number of failures of relative large open-
The results of the present study show that in caries-risk sandwich restorations due to fracture, compared to the 3-
patients, the failure rate of restorations was higher than in year results.2,12 One study on small closed-sandwich Class
non-risk patients. For the total-etch group, almost no failures I and II restorations reported a survival rate of 83% after 8
due to caries were seen in the low-risk group. Hence, the low years, which is in accordance with the sandwich restora-
annual failure rate of 0.8% that was found in this group in- tions in our present study. In that study as well, the majori-
dicates that the selected composite and adhesive system ty of failures was related to fracture and partial loss of the
are of high quality and both operators performed well. The restoration.15 A 17-year clinical trial on closed-sandwich
difference in annual failure rate between low-risk and high- restorations showed an acceptable survival rate, but the
risk patients and the finding that the main reasons for fail- most important reason for failure was fracture, while caries
ure did not appear before 3 to 4 years of clinical service (as was seldom found.8 It can be speculated that the restora-
expressed in Figs 2a and 2b) is remarkable. Today, many tions in that study were placed in patients with low caries
new composite resins and adhesive systems are put on the risk. Because no direct comparison was made in any of
market as an improvement of previous products. These new- these studies between sandwich- and total-etch restora-
er products are mostly tested in clinical studies carried out tions, the role of glass ionomers in the fracture behavior re-
in university clinics, largely with observation times less than mained unclear.
3 years. The moment that long-term data become available Based on the results of the present study, the elastic lay-
and materials are clinically proven, these materials are often er concept should be viewed with scepticism. It is more like-
classified as outdated by the manufacturers and replaced by ly that a shock-absorbing layer underneath the restoration
an “advanced formula”. facilitates the onset of restoration failure in the long run.
Recent clinical studies on two new tooth-colored materi- Whether this also occurs with other elastic layers, such as
als, Solitaire and Ariston, showed a considerable failure rate other glass-ionomer cements, thick adhesive resins, flow-
of 21% for Solitaire after 3 years and 17% for Ariston after 2 able composites or compomers, remains unclear and
years.14,19 Based on the results, the authors advised waiting should be investigated. Meanwhile, in the present study the
for at least the two-year clinical data before bringing new ma- most rigid combination of a thin adhesive resin with a hybrid
terials on the market. The results of the present study show composite resin showed excellent clinical performance over
that with materials that have been on the market for more a 10-year period.
than 15 years, excellent results can be obtained in general
dental practice. Most of the failures did not occur before 4
years of clinical service, and the survival of the restorations CONCLUSIONS
was significantly influenced by the caries risk level of the pa-
tients in the study. The present ADA guidelines require an 1. Total-etch restorations placed with a highly filled hybrid
18-month period of clinical service for acceptance of a new composite resin showed a higher clinical survival than
all-purpose composite resin. Based on the findings in the closed-sandwich restorations using a lining of RMGIC,
present study, the authors consider this period to be far too due to a lower fracture rate.
brief to detect possible shortcomings of dental materials, 2. Fracture and secondary caries, the most important rea-
and suggest a period of at least 4 years for evaluation of new sons for failure, mostly occurred after a period of more
materials. than three years.

474 The Journal of Adhesive Dentistry

 All C opyrig
eR

ht
ech
te

by
vo
rbe

Qu
Opdam et al
ha
nt lte
n

i
14. Ernst CP, Martin M, Stuff S, Willershausen B. Clinical performance of a
packable resin composite for posterior teeth after 3 years. Cline
3. Doubt should be cast on the alleged advantages of the
elastic layer underneath a composite resin restoration. 2001;5:148-155.
Oral
e nz
ssInvest
15. Gaengler P, Hoyer I, Montag R. Clinical evaluation of posterior composite
restorations: the 10-year report. J Adhes Dent 2001;3:185-194.
16. Hickel R, Manhart J. Longevity of restorations in posterior teeth and rea-
REFERENCES sons for failure. J Adhes Dent 2001;3:45-64.
17. Isokangas P, Alanen P, Tiekso J, The clinician's ability to identify caries risk
1. Abdalla AI, Davidson CL. Comparison of the marginal integrity of in vivo subjects without saliva tests—a pilot study. Com Dent Oral Epid 1993;21:8-
and in vitro Class II composite restorations. J Dent 1993;21:158-162. 10.
2. Andersson-Wenckert IE, van Dijken JW, Kieri C. Durability of extensive 18. Kemp-Scholte CM, Davidson CL. Marginal integrity related to bond strength
Class II open-sandwich restorations with a resin-modified glass ionomer and strain capacity of composite resin restorative systems. J Prosth Dent
cement after 6 years. Am J Dent 2004;17:43-50. 1990;64:658-664.
3. Ausiello P, De Gee AJ, Rengo S, Davidson CL. Fracture resistance of en- 19. Krämer N, Garcia-Godoy F, Frankenberger R. Evaluation of resin composite
dodontically-treated premolars adhesively restored. Am J Dent 1997; 10: materials. Part II: in vivo investigations. Am J Dent 2005;18:75-81.
237-241. 20. Opdam NJM, Bronkhorst EM, Roeters FJM, Loomans BAC. A retrospective
4. Ausiello P, Davidson CL, Cascone P, De Gee AJ, Rengo S. Debonding of ad- clinical study on longevity of posterior composite and amalgam restora-
hesively restored deep class II MOD restorations after functional loading. tions. Dent Mater 2007;23:2-8.
Am J Dent 1999;12:84-88. 21. Pallesen U, Qvist V. Composite resin fillings and inlays. An 11-year evalua-
5. Ausiello P, Apicalla A, Davidson CL. Effect of adhesive layer properties on tion. Clin Oral Invest 2003;7:71-79.
stress distribution in composite restorations – a 3D finite element analy- 22. Randall RC, Wilson NH. Glass-ionomer restoratives: a systematic review of
sis. Dent Mater 2002;18:295-303. a secondary caries treatment effect. J Dent Res 1999;78:628-637.
6. Baratieri LN, Ritter AV. Four-year clinical evaluation of posterior resin-based 23. Seppä L, H. Hausen H, L. Pöllänen L, K. Helasharju K, S. Karkkainen S.
composite restorations placed using the total-etch technique. J Esthet Past caries recording made in Public Dental Clinics as predictors of caries
Restor Dent 2001;13:50-57. prevalence in early adolescence. Com Dent Oral Epidem 1989;17:277-
7. Braga RR, Hilton TJ, Ferracane JL. Contraction stress of flowable composite 281.
materials and their efficacy as stress-relieving layers. J Am Dent Assoc 24. Tam LE, Pilliar RM. The effect of interface stiffness on dentin-composite in-
2003;134:721-728. terfacial fracture resistance. J Dent 2000;28:487-493.
8. Da Rosa Rodolpho PA, Cenci MS, Donassollo TA, Loguércio AD, Demarco 25. Vilkinis V, Horsted-Bindslev P, Baelum V. Two-year evaluation of class II
FF. A clinical evaluation of posterior composite restorations:17-year find- resin-modified glass ionomer cement/composite open sandwich and com-
ings. J Dent 2006;7:427-435. posite restorations. Clin Oral Invest 2000;4:133-139.
9. Davidson CL. Glass-ionomer bases under posterior composites. J Esthet 26. Wendt SL, Leinfelder KF. Clinical evaluation of Clearfil Photo Posterior: 3-
Dent 1994;6:223-226. year results. Am J Dent 1992;6:121-125.
10. De Munck J, Van Landuyt KL, Coutinho E, Poitevin A, Peumans M, Lam-
brechts P, Braem M, Van Meerbeek B. Fatigue resistance of dentin/com-
posite interfaces with an additional intermediate elastic layer. Eur J Oral
Sci 113 (1), 77–82.
11. Dietschi D, Kreji I. Adhesive restorations in posterior teeth: rationale for the
application of direct techniques. Oper Dent 2001;6:191-197. Clinical relevance: Posterior composite resin restorations
12. van Dijken JW, Kieri C, Carlen M. Longevity of extensive class II open-sand- placed with a closed-sandwich technique and a glass-
wich restorations with a resin-modified glass-ionomer cement. J Dent Res
1999;78:319-325. ionomer lining cement show more failure in the long term
13. van Dijken JW, Sunnegardh-Gronberg K. A four-year clinical evaluation of a than posterior composite restorations placed with a total-
highly filled hybrid resin composite in posterior cavities. J Adhes Dent etch technique.
2005;7:343-349.

Vol 9, No 5, 2007 475