d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939

available at www.sciencedirect.com

journal homepage: www.intl.elsevierhealth.com/journals/dema

A prospective 15-year evaluation of extensive

dentin–enamel-bonded pressed ceramic coverages

Jan W.V. van Dijken a,∗ , Lars Hasselrot b

a Faculty of Medicine, Department of Odontology, Dental School Umeå, Umeå University, SE-901 87 Umeå, Sweden
b Public Dental Health Center Korpen, Visby, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: Objectives. The purpose of this study was to investigate the durability of extensive
Received 28 September 2009 dentin–enamel-bonded posterior ceramic coverages in a 15 years follow-up.
Received in revised form Methods. All extensive dentin–enamel-bonded posterior partial and complete all-ceramic
9 April 2010 coverages placed during the period November 1992–December 1998 were included. In 121
Accepted 25 May 2010 patients, 252 coverages (IPS Empress) were placed. The adhesive bonding to dentin and
enamel was performed with three 3-step and one 2-step etch and rinse bonding. In 106
restorations the classic Syntac was used in combination with the dual-cured resin composite
Keywords: Variolink. The other restorations were luted with the chemically cured resin composite Bisfil
Cement 2B and bonded with 3-step etch and rinse systems, classic Gluma (37), Allbond 2 (57), Syntac
Ceramic (32) or the 2-step etch and rinse system, One step (20). The ceramics were evaluated yearly
Composite by modified USPHS criteria during 15 years.
Clinical Results. Postoperative sensitivity was registered in 4 patients during bite forces lasting for
Resin 2–4 weeks. Fifty-five of 228 coverages (24.1%) failed. The mean observation period of the
Bonding agents acceptable coverages was 12.6 years (range 11–15 years). The main reasons for failure were
Dental ceramics lost restorations (18), ceramic fracture (16), and secondary caries (11). Significant differences
Degradation in failure rate were observed between the dentin bonding agents but not between the two
luting agents. Ceramic coverages placed on non-vital teeth failed in 39% and on vital teeth
in 20.9% (p = 0.014). Logistic regression indicated three significant predictors for failure of the
coverages: gender and parafunctional habits of the patient and non-vitality of the tooth.
Significance. The technique investigated showed advantages like less destruction of healthy
tissue, and avoiding of endodontic treatment and/or deep cervical placement of restoration
margins to obtain retention.
© 2010 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.

dental ceramic systems have been used during the last

1. Introduction
The popularity of tooth-colored posterior restorations has
increased during the last years because of a growing demand
for esthetics and concern about the biocompatibility of amal-
gam [1]. Applying the adhesive bonding concept, several
decades for veneer, inlay and onlay restorations [2]. Most
ceramic materials can be bonded to the underlying condi-
tioned tooth surface after etching with hydrofluoric acid or
ammonium bifluorid, mediated by use of an enamel/dentin
bonding system and a resin composite luting material. Acid
etching increase surface roughness and wetting of enamel,

∗
Corresponding author. Tel.: +46 90 7856034/7856047; fax: +46 90 770580.
E-mail address: Jan.van.Dijken@odont.umu.se (J.W.V. van Dijken).
0109-5641/$ – see front matter © 2010 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.dental.2010.05.008
930 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939
dentin and ceramic surfaces, which promote mechanical
interlocking of resin bonding systems.
Bulk fracture and loss of restoration have been reported
as the main reasons of failure in short-term evaluations of
inlays and onlays [3]. Leucite-reinforced glass–ceramic inlays
showed improved clinical durability compared to fired ceram-
ics [3–5]. The advantage of bonded ceramics can be expected
in the extremely non-retentive coverage situation [6–9]. Tra-
ditional crown preparation techniques will in these cases
occasionally result in extensive preparation and/or endodon-
tic treatment in combination with a post and core placement
in order to obtain retention. The minimal preparation for the
bonded ceramic is less traumatic for the tooth, and pulp vital-
ity can be preserved [9]. There are few preparation design
standards for the ideal dentin–enamel-bonded crowns, and
Burke [10] showed in vitro no differences between varying
degrees of tooth preparation to enhance protection to frac-
ture. The durability of the bonded ceramic will depend on the
strength of the bond between tooth, luting system and ceramic
and on the inherent strength of the ceramic. The use of chem-
ically cured resin composite cements have been suggested
to obtain optimal conversion and decreased stress formation
during polymerization [5,11].
Few studies report the longevity of dentin–enamel-bonded
all-ceramic crowns also defined as “partial or full cover-
age restorations in which an all-ceramic is bonded to the
underlying dentin and any available enamel using a resin
luting material” [1,9,12–14]. In a systematic review assessing
the 5-year survival rates of single crowns, densily sin-
tered aluminum crowns and reinforced glass–ceramic crowns
(Empress) showed survival rates comparable to those seen
for porcelain-fused-to-metal (PFM) crowns [15]. The short-
term results of the extensive adhesively luted coverages are
promising but clinical long-term data are not available [9]. The
aim of this study was to investigate the long-term durability
of these extensive dentin–enamel-bonded posterior ceramic
coverages. In addition the effect of luting cement, bonding
system and preparation type was studied. The hypothesis
tested was that there was (1) no difference in durability for
the ceramic coverage placed with different luting agents and
different bonding agents, and (2) no difference in durability
between vital and endodontic treated teeth.
2. Materials and methods
During the period November 1992–December 1998, 262
enamel–dentin-bonded ceramic partial and complete cover-
ages (IPS Empress, Ivoclar, Schaan, Liechtenstein) were placed
by four operators in 121 patients, 75 women and 46 men with a
mean age of 52 years (range 26–81). The study was randomized
in a way that all patients, which did need an extensive poste-
rior all-ceramic restoration, during this period, were included
in the study after giving their agreement. No patient or patient
group was excluded. The participants were treated on a regu-
lar basis at the operator’s Public Dental Health Service clinics
in Umeå and Visby, Sweden or the Umeå dental school clinic.
The study was approved by the local ethics committee at the
University of Umeå. The socio-economic status of the partici-
pants was varying from low to high. The reasons for placement
were filling- or tooth fracture, secondary caries and replace-
ment of extensive amalgam restorations. Preparations with
retention obtained as for conventional metallic crowns by par-
allel walls and increased height of the walls were not included.
The evaluated ceramics were distributed as follows: in the
lower arch 20 premolars and 120 molars, in the upper arch 44
premolars and 68 molars. All teeth were in occlusion and had
at least one proximal contact. The cavities were prepared with
butt-joint margins using tapered diamond burs. Sharp internal
and external preparation angles were rounded to assure opti-
mum strength of the ceramic. The occlusal thickness of the
coverage was at least 1.5 mm. To increase the enamel surface
of the preparations minimal shoulder or chamfer preparations
were made in part of the teeth in case of enough thickness of
the remaining cusp(s). The thickness of the remaining cusp
was estimated by the operator to be at risk for fracture or
not before a preparation was performed. In case of risk, the
remaining cusp(s) were left unprepared. The different forms
of preparations evaluated have been reported earlier [9]. Prepa-
ration group 1: partial coverage with no shoulder preparation
of thin remaining cusp wall (premolars, n = 4; molars, n = 25);
Preparation group 2: minimal retentive partial coverage prepa-
ration, with a combination of unprepared thin remaining cusp
wall(s), and cusps which have been prepared with a shoulder
or chamfer (premolars (with one remaining unprepared cusp
wall), n = 18; molars (with one or two remaining unprepared
cusp walls), n = 70); Preparation group 3: minimal retentive
full coverage crowns with shoulder or chamfer preparation of
all remaining cusps (premolars, n = 21; molars, n = 63); Prepa-
ration group 4: non-retentive endodontic treated teeth, no
post and core treatment (premolars, n = 13; molars, n = 14). The
numbers given are excluded the drop out restorations. Under-
cuts were blocked with resin-modified glass ionomer cement
or dentin-bonded resin composite material or removed by
preparation. A full arch impression was made with a custom-
made acrylic resin tray with polyvinyl-siloxane impression
materials (President, Coltène, Altstätten, Switzerland; Provil,
Bayer, Leverkussen, Germany; or similar materials). Tempo-
rary restorations were cemented with an eugenol-free cement
(Fermit, Vivadent, Schaan, Liechtenstein; Tempbond-NE, Kerr,
Karlsruhe, Germany), or phosphate cement was used to cover
the non-retentive preparations. All-ceramics were processed
by four experienced technicians. The internal surfaces of the
ceramic coverages were etched with hydrofluoric acid in the
laboratory. At the second appointment, the operation field
was isolated with cotton rolls and a conventional saliva suc-
tion device or rubber dam. At the try-in of the coverages,
anatomical form, marginal adaptation and color were eval-
uated. After this the fitting surface of the ceramic restoration
was cleaned by a 2–3 s etching with 9.5% hydrofluoric acid or
20 s etching with 36% phosphoric acid to remove any possible
contaminants during the try-in like saliva or crevicular fluid.
The etching times were based on internal SEM observations
of cleaning methods of saliva contaminated with hydrofluo-
ric acid pre-etched ceramic surfaces. A silan coupling agent
(Ultradent, Salt lake City, UT, USA; Monobond S, Vivadent) was
then applied for 2 min. The teeth were etched with 36% phos-
phoric acid for 15 s, enamel 10 s followed by an enamel and
dentin etch for another 5 s. The 15 s etching time for enamel
has been proven to give good longevity in our long-term eval-
 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939
Table 1 – Distribution of the adhesive systems for the
preparation groups.
Bonding Total
Gluma Allbond 2 Syntac One step
Preparation group
1 1 7 11 10 29
2 10 16 58 4 88
3 20 20 39 5 84
4 2 10 14 1 27
Total 33 53 122 20 228
uations, while the 5 s dentin etch is considered to optimize
penetration of the demineralized dentin and minimize the
risk for nanoleakage. The cavity was thoroughly rinsed with
water for 20 s, and carefully dried taking care not to dry out
the surface (wet technique introduced by J. Kanca). Four dif-
ferent dentin bonding systems used were applied according
to the manufacturerı̌s instructions. In 106 restorations the
classic Syntac adhesive system (Vivadent) was used in combi-
nation with a dual-cured resin composite (Variolink, Vivadent)
systematically by one of the operators. In the other center
restorations were luted with a chemically cured resin com-
posite (Bisfil 2B, Bisco Inc., Itasca IL, USA). Three 3-step etch
and rinse systems were used systematically in different time
periods by the other operators. 37 were bonded with the Gluma
system (original Gluma, Bayer, Dormhagen, Germany), 57 with
the Allbond 2 system (Allbond 2, Bisco Inc.), 32 with the classic
Syntac system (Vivadent). Another 20 coverages were placed
at the end of the study with the simplified 2-step etch and
rinse system One step (Bisco Inc.) as a comparison with the
3-step system of the same manufacturer. The distribution of
the adhesive systems for the different preparation groups is
shown in Table 1. The luting materials were applied to the
fitting surface of the coverages. Excess cement was removed
immediately after placement with probe or brush-tips and
proximal floss. The dual-cured agent was light cured from all
directions for 40–60 s per location (VRC 400, Dentronic, USA;
1000 mW/cm2 ).
Occlusion was checked and adjusted as necessary after
placement based on a dynamic occlusion. The reconstruc-
tions were finished with diamond or carbide finishing burs,
polishing stones and the Profin finishing system (Dentatus,
Hägersten, Sweden) under water cooling followed by polishing
stones.
2.1. Evaluation
Each coverage was evaluated after insertion (baseline) and
then every year during the follow-up regularly by two cal-
ibrated investigators. Evaluations at the Visby clinic were
performed by the operator and at the end of the study by
two other investigators. All were well familiar with the eval-
uation system. Disagreement was resolved by consensus. A
slight modification of the United States Public Health Sys-
tem (USPHS) criteria was used to evaluate the quality of the
restorations (Table 2; [9]). At the yearly recalls, none of the
evaluators did know the adhesive luting technique used for
the restorations to be evaluated.
931
Postoperative sensitivity was analyzed according with the
system used by Borgmeijer et al. [16]. Bite-wing radiographs
were taken regularly and in addition color slides were made
of selected cases.
The follow-up was considered as a drop out when a partic-
ipant was not evaluated during 3 years. The restorations were
recorded as in function, censored if the participants dropped
out, or failed if they were repaired or replaced or the tooth was
indicated for extraction.
2.2. Statistical analysis
The SPSS (Statistical Package for the Social Sciences; SPSS
Inc., Chicago, IL, USA) version 17.1 was used to process the
data. The characteristics of the restorations were described
by descriptive statistics using frequency distributions of the
scores. The number of failed restorations observed during the
15 years and the survival age of all-ceramics vs. gender, age of
the participants, luting agent and bonding system, vitality of
the teeth, preparation type and modified Ryge criteria scores
at the final recalls were tested with non-parametric tests
(Kruskal–Wallis, Mann–Whitney, Monte Carlo exact test, uni-
variate analysis of variance test and post hoc Bonferrini test)
[17]. Kaplan–Meier was used to describe survival functions.
To determine the individual contribution of the different fac-
tors to predict the outcome of the bonded coverage, a logistic
regression was performed. As dependent variable the failure of
the coverage and as predictor variables all factors which could
influence the outcome with the 5- and 10-year evaluations as
endpoint: gender, parafunctional habits, jaw, tooth type, vital-
ity of the tooth, preparation type, luting agent, bonding agent.
The null hypothesis was rejected at 5% level.
3. Results
During the follow-up 16 participants, 7 women and 9 men with
24 ceramic coverages in 19 molars and 7 premolar teeth, were
not able to continue the evaluations due to moving or death.
The other 228 Empress coverages, 146 in women and 82 in
men, were followed during the whole follow-up. Postopera-
tive sensitivity was registered in 4 patients during bite forces
lasting for the first 2–4 weeks. During the 15 years follow-
up, 55 non-acceptable coverages (24.1%) were observed. The
number of failed restorations, reasons for failure and fail-
ure year are shown in Table 3. The mean (SD) and median
(minimum–maximum) of years in service for all evaluated
ceramics was: 11.8 (3.6) and 12.5 (2–15). These values for all
acceptable ceramics and all failed ceramics were 12.6 (1.7) and
13.5 (11–15), respectively, 5.7 (3.0) and 6.0 (2–13). Small chip
fractures were observed in 5 other restorations. Kaplan–Meier
survival analysis of the 228 restorations evaluated as a func-
tion of gender, tooth vitality, luting agent, bonding system
and preparation groups are shown in Figs. 1–5. Men showed
a 31.7% failure frequency and women 19.9% (p = 0.045; Fig. 1).
The cumulative failure frequencies for premolars was 26.8%
and for molars 25% (p = 0.59), for upper jaw ceramic coverages
22% and for lower jaw ones 25% (p = 0.89). The failure frequency
in vital teeth was 20.9% (39/187) and in non-vital teeth 39.0%
(16/41) (Fig. 2). The differences were significant both for failure
932 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939

Table 2 – Slightly modified USPHS criteria according to van Dijken et al. [4].
Category Score Criteria

Acceptable Unacceptable
Anatomical form 0 The restoration is continuous with tooth anatomy
1 Slightly under- or over-contoured restoration; marginal
ridges slightly undercontoured; contact slightly open
(may be self-correcting); occlusal height reduced locally
2 Restoration is undercontoured, dentin or base exposed;
contact is faulty, not self-correcting; occlusal height
reduced; occlusion affected
3 Restoration is missing partially or totally; fracture of
tooth structure; shows traumatic occlusion; restoration
causes pain in tooth or adjacent tissue

Marginal adaptation 0 Restoration is continuous with existing anatomic form,

explorer does not catch
1 Explorer catches, no crevice is visible into which
explorer will penetrate
2 Crevice at margin, enamel exposed
3 Obvious crevice at margin, dentin or base exposed
4 Restoration mobile, fractured or missing

Color match 0 Excellent color match

1 Good color match
2 Slight mismatch in color, shade or translucency
3 Obvious mismatch, outside the normal range
4 Gross mismatch

Marginal discoloration 0 No discoloration evident

1 Slight staining, can be polished away
2 Obvious staining cannot be polished away
3 Gross staining

Surface roughness 0 Smooth surface

1 Slightly rough or pitted
2 Rough, cannot be refinished
3 Surface deeply pitted, irregular grooves

Caries 0 No evidence of caries contiguous with the margin of the

restoration
1 Caries is evident contiguous with the margin of the
restoration

frequency (p = 0.014) and for years of survival (p = 0.033). The
relative cumulative failure frequency for the chemical-cured
resin composite luted ceramics was 24.1% and for the dual-
cured resin composite luted coverages 24.4% (p = 0.55; Fig. 3).
The relative cumulative failure frequencies for the coverages
bonded with the four bonding systems were: Gluma: 27.3%;
Allbond 2: 22.6%; Syntac: 20.5% and One step: 45.0% (Fig. 4).
One step differed significantly compared to Syntac (p = 0.02)
and Allbond 2 (p = 0.0001).
The four preparation groups were equally distributed
between premolars and molars (p = 0.41). The relative cumu-
lative failure frequencies for the four preparation groups
were: group 1: 34.5%; group 2: 18.2%; group 3: 22.6% and
group 4: 37.0% (Fig. 5). A significant difference was observed

Table 3 – Reasons and number of failures at the recall years for the ceramic coverages.
Failure mode Failure year Total

2 3 4 5 6 7 8 9 10 11 12 15
Lost 4 3 3 1 1 2 3 1 18
Ceramic fracture 2 1 4 2 4 1 1 1 16
Caries 2 3 2 2 1 1 11
Endodontic reasons 1 2 3
Extraction periodontal reasons 1 2 3
Crown fracture 1 1
Root fracture 1 1 1 3

Total 8 6 7 6 2 10 7 1 4 2 1 1 55
 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939 933

Fig. 1 – Kaplan–Meier survival analysis of the 228 ceramic Fig. 2 – Kaplan–Meier survival analysis of the 228 ceramic
coverages evaluated as a function of gender of the coverages evaluated as a function of tooth vitality, vital
participants. teeth vs. root-filled teeth.

between the groups when tested against years of survival of

the included coverages (p = 0.01; post hoc Bonferrini test: group
3 vs. group 4: p = 0.029).
The Kaplan–Meier estimates and 95% confidence intervals
of the variables analyzed are shown in Table 4. The scores for
the evaluated clinical modified Ryge criteria for all restora-
tions at baseline (year 0) and the last recall for the respective
ceramics are shown in Table 5. Nagelkerke R2 at 5 years and 10
years were 0.202 and 0.154, respectively. Gender and parafunc-
tioning (bruxing) were the two significant variables in Wald
statistics at 5-year with exp(B) of 3.074 and 0.272, respectively.
At 10 years, there were three significant variables gender, vital-
ity and parafunctioning, with exp(B) of 1.959, 0.415 and 0.383,
respectively (Tables 6 and 7). The scores of the last recall eval-
uations performed with the modified Ryge criteria differed
significantly with the baseline ones for all variables except for
surface roughness.

Table 4 – Kaplan–Meier estimates (mean) and 95% confidence intervals for gender, premolar vs. molar teeth, preparation
groups, vital vs. non-vital teeth, luting agents and bonding systems.
Estimate (mean) SD 95% Confidence interval
Gender
Men 11.98 0.55 10.81–12.95
Women 13.17 0.32 12.55–13.79

Premolar vs. molar

Premolar 12.69 0.60 11.52–13.86
Molar 12.72 0.32 12.08–13.35

Preparation group
Group 1 10.76 0.86 9.07–12.44
Group 2 13.31 0.42 12.48–14.14
Group 3 12.96 0.43 12.12–13.80
Group 4 11.28 0.95 9.41–13.14

Tooth vitality
Vital teeth 12.97 0.31 12.37–13.57
Non-vital teeth 11.48 0.72 10.07–12.89

Luting agents
Bisfil 2B 12.50 0.39 11.74–13.26
Variolink 13.00 0.41 12.20–13.81

Bonding systems
Gluma 12.67 0.77 11.16–14.17
Allbond 2 12.68 0.60 11.50–13.86
Syntac 13.15 0.34 12.48–13.83
One step 9.57 1.13 7.35–11.79

Overall 12.71 0.29 12.15–13.27

934 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939

Fig. 3 – Kaplan–Meier survival analysis of the 228 ceramic Fig. 4 – Kaplan–Meier survival analysis of the 228 ceramic
coverages evaluated as a function of luting agent used, coverages evaluated as a function of bonding system used:
chemical-cured resin composite (Bisfil 2B) vs. dual-cured 3-step etch and rinse Gluma, Allbond 2 and Syntac and
resin composite (Variolink). 2-step etch and rinse One step.

soft tissues contiguous the metallic framework. Alternative

4. Discussion
Traditionally, metal and PFM have been used for the treatment
of severely damaged posterior teeth which are in need of one
or more cusp coverages. They limit tooth fracture and main-
tain a relatively long oral durability [18]. In two meta-analyses
the survival rate of PFM after 15 years was between 69% [19]
and 74% [15]. However, a disadvantage of PFM crowns is the
substantial removal of sound tooth tissue requested to obtain
sufficient retention, and an unnatural grayish appearance of
treatments for severely damaged teeth are the placement
of extensive amalgam, resin composite or ceramic restora-
tions [6–9,20]. Patients demands for tooth-colored restorations
increased in many countries during the last 15 years, not at
least because the growing concerns about amalgam. During
the nineties, reports indicated that direct resin composites
restorations had disadvantages like setting stress, lack of cer-
vical marginal adaptation, limited strength and color stability.
Therefore, many practitioners were discouraged to make pos-
terior resin composite restorations. They chose all-ceramic

Table 5 – The scores for evaluated criteria of the partial- and total ceramic coverages, given as relative frequencies (%), at
baseline (B) and at the last recall of the follow-up (E). Baseline vs. end evaluation tested with Friedman test.
0 1 2 3 4 Significance
a
Anatomical form
B 83.8 16.2 0 0 p = 0.003
E 75.7 8.1 0.9 15.3

Marginal adaptationa
B 60.5 39.0 0.5 0 0 p = 0.000
E 27.8 48.4 6.7 0.9 16.1

Color match
B 37.7 50.4 11.4 0.4 0 p = 0.000
E 17.0 62.8 18.1 2.1 0

Marginal discoloration
B 99.6 0.4 0 0 p = 0.000
E 61.2 26.6 11.7 0.5

Surface roughness
B 62.7 37.3 0 0 p = 0.275
E 58.5 33.5 8.0 0

Cariesa
B 100.0 0 p = 0.000
E 93.1 6.9
a
Cumulative relative frequencies.
 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939
Fig. 5 – Kaplan–Meier survival analysis of the 228 ceramic
coverages evaluated as a function of preparation groups.
restorations, veneers and inlays, as a main restorative choice,
despite missing evidence. Dental ceramics are brittle materi-
als, with relatively high compressive strengths but low tensile
and flexural strengths and fracture toughness. In case of
extensively restored teeth or with short crown height, these
fragile restorations are to a very high degree depended on suf-
ficient support and bond to the underlying tooth substance.
The introduction of amphiphilic bonding systems allowed
a strong adhesive bond of the, by hydrofluoric acid etched,
all-ceramic restoration with the prepared teeth via an inter-
mediate resinous luting cement layer [9,21]. Non-acid-etched
all-ceramic coverages showed a 2.2 times greater risk of fail-
ure compared to the adhesively bonded, acid etched, ones.
A 76% overall probability of survival was observed for acid-
etched Dicor crowns after 14 years on teeth with adequate
tooth structure and in patients with good oral hygiene [21].
The probability of survival for molars was 48.8%. The use of
adhesively bonded all-ceramic coverages in severely damaged
teeth has increased dramatically during recent years. How-
ever, even if much literature is available about the longevity
and clinical performance of inlays and crowns, data of adhe-
sively bonded all-ceramic partial and total coverages are rare.
There is still a lack of evidence for their use.
The clinical performance of adhesively bonded all-ceramic
restorations has mostly been studied in short-term studies
for ceramic inlays (1–5 years), while only a few extended the
observation time of 10 years. Most of these longer evaluations,
studied the CAD/CAM Cerec system using different prefabri-
cated ceramic ingots [11,14,22]. Reported survival probabilities
of these ceramic inlays after 10 years were around 90%. A 16
years follow-up of Cerec inlays reported a success rate of 84%
[22]. The more brittle feltsphatic ceramic inlays showed dis-
tinctly lower survival probabilities of 13–61% after observation
periods of 6–7 years [4,12,13,23]. For the with leucite crys-
tals reinforced pressed all-ceramic inlays only one prospective
study extended 10 years reporting a success rate of 84% after
12 years [14]. A few of the published inlay studies included
also small number of (the larger) onlay restorations. However,
935
the low number restorations evaluated made statistical calcu-
lations and clinical conclusions difficult [11,12,14,24,25]. The
long-term evidence for partial and total adhesively bonded
all-ceramic coverages is unfortunately rather limited in terms
of the follow-up period. A large difference in survival was
observed in a 7-year evaluation between a fluorine contain-
ing tetra silicate glass–ceramic (Dicor) and leucite-reinforced
pressed ceramic (Empress I) partial coverages. Cumulative
survival rates of 40% for Dicor and 5.3% for Empress, were
reported, however only 35 and 19 restorations, respectively,
were evaluated [12]. In another 7-year follow-up of the same
group of 42 partial Empress coverages, luted with four different
resin luting agents, an 81% survival was reported. Smales and
Etemadi [13] observed a 26.9% failure rate after 6 years for pos-
terior sintered feldsphatic onlays. Fradeani and Redemagni
[24] reported a probability of survival of 84% for 32 posterior all-
ceramic pressed crowns after a 4–11 years follow-up. The value
of studies including only small number of coverages found in
many prospective follow-up studies is questionable [12,24–27].
A recent systematic review of ceramic inlays concluded that
the clinical effectiveness of ceramic inlays cannot be deter-
mined from studies which included relatively small numbers
of restorations with relatively short-term evaluations over 1–5
years [28].
The loss of patients in prospective longitudinal studies is
always a limitation. The drop out in our study was 9.5% at
15 years. This can be compared to drop out figures of other
long-term follow-ups of ceramic inlays of 25.7% at 10 years
[11], and 40% after 11 and 12 years [14,24]. In contrast to the
present study, many prospective all-ceramic studies were per-
formed in selected patient groups, excluding in many cases
patients with non-optimal oral hygiene and bruxing habits
[14,26,27,29–31], caries risk patients [24,32] or cases when no
rubberdam could be applied [14,29]. None of the retrospective
studies of all-ceramics did report exclusion criteria. Exclud-
ing of different risk groups imply that the observed results
for the respective selected patient groups cannot directly be
transferred to the general practice patients including all risk
groups.
In most of the studies evaluating all-ceramic crowns, the
crowns were placed on traditional macro-mechanical reten-
tive preparations, including cervical deep preparations with a
circumferential shoulder. This extensive way of preparation to
obtain macro-retention is in large contrast to the partial and
total coverages described. Lehner et al. [33] reported a failure
rate of 11.6% for 138 adhesively luted IPS Empress full crowns
after 5 years. They prepared with a circular shoulder, to get
additional traditional retention.
Suggested preparation designs for the dentin–enamel-
bonded partial coverages have in some studies been based on
traditional crown constructions, while others suggested that
the traditional rules and principles of precision prosthodon-
tics are no longer applicable. The longevity of this has not been
determined [34]. In traditional prosthetic preparation tech-
niques a reducing of the cusp and onlay technique have been
recommended to decrease unfavorable cusp fractures. A cus-
pal overlaying should allow a material thickness of preferably
2 mm. By means of adhesive techniques, cuspal reinforce-
ment provided retention but also enhanced fracture resistance
and sealing of dentin-bonded crowns in vitro [10], which was
936 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939

Table 6 – Logistic regression analysis of variables which significantly predict failure of the coverage at five years: gender
and parafunctional habits (bruxism) of the participants.
Variables in the equation

B S.E. Wald df Sig. Exp(B) 95.0% C.I. for exp(B)

Lower Upper
a
Step 1
Bruxism (1) −1.262 0.423 8.892 1 0.003 0.283 0.123 0.649
Constant −1.149 0.318 13.022 1 0.000 0.317

Step 2b
Gender (1) 1.123 0.432 6.770 1 0.009 3.074 1.319 7.161
Bruxism (1) −1.301 0.434 8.975 1 0.003 0.272 0.116 0.638
Constant −1.639 0.393 17.394 1 0.000 0.194
a
Variable(s) entered on step 1: bruxism.
b
Variable(s) entered on step 2: gender.

confirmed in the present study. Only one cusp fracture was
observed during the follow-up. The preparation group without
shoulder preparations on remaining cusps (group 1) showed
however, a 34.5% failure rate, which was higher than the over-
all failure rate and almost as high as the non-vital teeth group.
Cross-sectional evaluations of moderate sized cavities with
filling materials have been in favor of amalgam, while multi-
surface restorations showed decreased durability for resin
composites [1,35]. Van Nieuwenhuysen et al. [20] reported, in
a follow-up of similar length as the present study, but with
a limited amount molars involved, for extensive posterior
restorations (crowns, amalgam and resin composite) a 28%
cumulative number of failures. This can be compared with
the 24.5% failure rate in the present study. Their Kaplan–Meier
median survival rates were for amalgam restorations 12.8
years, for resin composites 7.8 years and for crowns when
repairs as well as replacements and extractions were treated
as failures 14.6 years, while we observed a Kaplan–Meier esti-
mate of 12.7 years. The rates for amalgam and crowns were
comparable with the results of the adhesive luted all-ceramic
coverages as mentioned above. One has to observe that the
teeth involved in the present study showed extreme loss of
tooth substance while a high frequency of molar teeth inves-
tigated (78%).
The clinical durability and fracture resistance of dentin-
bonded ceramic restorations is strongly dependent on the
time-dependent degradation of the ceramic material, luting-
and bonding agent [9,10]. It has to rely on the quality and
durability of the micromechanical bond to enamel and dentin,
which allow minimal tooth tissue removal and less pulpal
irritation [9]. The biomaterial-tooth interfaces are despite
good initial bond strength subjected to mechanical as well
as chemical degradation. Mechanically by occlusal forces and
chemically by hydrolytic degradation of the exposed collagen
fibrils and plasticizing of the polymer matrix. A reduction of
the bond durability is the result of the ingression of water.
Long-term evaluations of the durability of the dentin bond
strength of different adhesive systems in non-carious cervi-
cal lesions showed a continuous increasing frequency of lost
of retention [36,37]. For the adhesively bonded all-ceramic

Table 7 – Logistic regression analysis of variables which significantly predict failure of the coverage at ten years: gender
and parafunctional habits (bruxism) of the participants and vitality of the tooth.
Variables in the equation

B S.E. Wald df Sig. Exp(B) 95.0% C.I. for exp(B)

Lower Upper
a
Step 1
Bruxism (1) −0.926 0.349 7.055 1 0.008 0.396 0.200 0.784
Constant −0.501 0.283 3.123 1 0.077 0.606
Step 2b
Vitality (1) −0.831 0.395 4.434 1 0.035 0.436 0.201 0.944
Bruxism (1) −0.965 0.354 7.416 1 0.006 0.381 0.190 0.763
Constant 0.181 0.431 0.177 1 0.674 1.199
Step 3c
Gender (1) 0.672 0.343 3.838 1 0.050 1.959 1.000 3.838
Vitality (1) −0.879 0.399 4.850 1 0.028 0.415 0.190 0.908
Bruxistm (1) −0.960 0.358 7.205 1 0.007 0.383 0.190 0.772
Constant −0.044 0.449 0.009 1 0.923 0.957
a
Variable(s) entered on step 1: bruxism.
b
Variable(s) entered on step 2: vitality.
c
Variable(s) entered on step 3: gender.
 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939
restorations an increased frequency fractured and lost ceram-
ics by time may therefore be expected. Evaluating the failure
time pattern as shown in Table 2, the highest frequency of
ceramic fractures and lost restorations failures occurred dur-
ing the first 8 years and thereafter continuing in a decreasing
frequency. The effectiveness and quality of dual-cured resin
composite luting agents in adhesive dentistry have been ques-
tioned [11]. Ceramics with a thickness more then 3 mm limits
the transfer of light energy, resulting in a non-optimal con-
version rate of the dual-cured cement. The lower conversion
rate in the deeper parts of the cement layer was, however, not
expressed by a significant higher failure rate for the restora-
tions luted with the dual resin composite. The first hypothesis
was therefore accepted concerning the luting agents. Three
of the dentin bonding systems evaluated in the present study
represented the 3-step amphiphilic systems with different sol-
vents, while the fourth was a simplified 2-step etch and rinse
version of one of the 3-step etch and rinse systems (Allbond
2). They contained acetone and ethanol, acetone and water or
water only as transport agent to penetrate the demineralized
dentin and create a molecular intertanglement network with
the collagen fibrils. Clinical consequences of a failed bond can
be several: loss of restoration or fracture of the ceramic due
to loss of retention area, marginal discoloration, hypersensi-
tivity and/or secondary caries. The total failure rates for the
bonding systems varied between 20.5 and 45.0%. A significant
higher failure rate was found for the simplified system and the
first hypothesis was therefore rejected concerning the bonding
systems evaluated. The inferior bond to dentin of the simpli-
fied bonding was confirmed in an earlier clinical follow-up in
cervical non-carious lesions [38].
The most frequent reason for failure in non-ceramic
restorations has been reported to be secondary caries, frac-
ture of the restoration and fracture of the cusp in partial
restorations [1,20]. The predominant reason for failure of
all-ceramic inlays was bulk fracture [3,9,14]. In a review of all-
ceramic crowns, crown fracture was the most common reason
strongly related to the location with a 1:3 ratio for premo-
lars vs. molars [39]. The studied ceramic coverages mainly
failed also because of loss of retention and material frac-
ture. The third common failure reason was secondary caries.
The adhesive bonding of restorations to dentin has been con-
sidered to be weak and bonding to cervical dentin margins
has been suggested to be a contraindication. The majority
of the restorations in this study had parts of the cervical
margins located in dentin. In accordance to findings of other
recently published studies of adhesive restorations only few of
the ceramic restorations showed secondary caries (11) during
the 15 years evaluation. All lesions were located in proxi-
mal cervical locations, many diagnosed as surface lesions.
The majority of the caries lesions were found in caries risk
patients. The low frequency secondary caries in adhesively
bonded all-ceramic restorations has been observed in sev-
eral studies [2,4,9,11,14,22,23,29,30]. Tooth fracture has been
reported to be the main reason for failure for treatment of
extensive amalgam restorations [8], which was in large con-
trast with the present low frequency failure rate observed (only
one case).
No difference was found in failure behavior and rate
between molars and premolars which was in contrast to some
937
other all-ceramic crown studies [21,22]. This may partly be
due to that in the present study also the premolar teeth were
severely damaged compared to those evaluated earlier. Differ-
ence in durability between men and women has very seldom
been reported. We found a significant lower survival rate for
men with a higher failure frequency by loss of retention and
caries, as also shown by others [40]. A higher frequency partic-
ipant with parafunctional habits and caries risk was observed
in the male group. No difference was observed between the
survival of restorations in the maxilla and mandibula as also
shown in earlier studies for metallic crowns [41]. A higher
failure rate for root filled compared to vital tooth has been
demonstrated earlier for metallic and PFM crowns [18,41].
In the present study we found also a significant higher fail-
ure rate for endodontically treated teeth as earlier shown
for ceramic inlays [42]. The second hypothesis was there-
fore rejected. It can be speculated that root-filled teeth would
have a more reduced retention area compared to vital teeth.
The teeth in group 3, including the most severe damaged
teeth, showed a significant higher success rate than the non-
vital teeth in group 4. Another more realistic explanation
may be found in the differences in substrate to which the
hydrophilic primers were applied, hydrophilic dentin in vital
teeth vs. more sclerotic less-water containing dentin tissue in
endodontically treated teeth. Van Nieuwenhuysen et al. [20]
showed also that vitality of the teeth was a significant vari-
able for the relative failures. In the multiple regression analysis
performed, vitality of the tooth was one of three significant
predictors of the outcome of the bonded coverages at 10 years,
indicating a higher probability of bond failure for the non-vital
endodontic treated tooth after a long intra-oral period. It is
interesting that the two other and higher significant predic-
tors at both 5 and 10 years, gender and parafunctioning, are
patient related and not material or tooth related factors.
During the first years of use of adhesively bonded ceramics,
it was stated that abrasion-erosion of the luting agent in the
oral environment was a high failure risk by time. The resulting
so called ditching occurred in the majority of cases during the
first years especially in occlusal margins. Ditching has been
shown to be self-limiting and did not increase during the fol-
lowing years and did not result in clinical problems [2,4,9].
The tooth saving preparation used for the studied cover-
ages minimized the risk for pulpal complications as shown
by the low number of teeth requiring endodontic treatment
during the long follow-up, while only four teeth showed
uncomplicated postoperative sensitivity symptoms at base-
line. The number of failures obtained in this study have to
be weighed against the time and risks involved in performing
endodontic treatment and preparation for traditional crown
therapy to obtain adequate resistance and retention. In case
traditional full crown treatments would have been performed,
endodontic treatment should have been necessary in many of
the experimental teeth in order to find macro-mechanically
retention with a post and core placement. Other advantages
of the investigated technique were good esthetics and a higher
frequency of supra-gingival placed cervical margins with good
adaptation. The number of failures should therefore be con-
sidered as a relative number and not as absolute. The fact
that most failures observed, like fractures of the ceramic or
lost restorations, were easy to repair, without more destruc-
938 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939
tion of healthy tissue, expresses the conservative option of the
technique.
5. Conclusions
It can be concluded that the longevity of the dentin–enamel-
bonded coverages during this study indicate a reduced need
for traditional full coverage therapy in teeth with extensive
loss of tooth substance. The technique investigated showed
advantages like less destruction of healthy tissue, avoiding
of endodontic treatment and/or deep cervical placement of
restoration margins to obtain retention, and good esthetics.
Acknowledgements
This study was partly supported by the Swedish Medical
Research Council, the Swedish Dental Association and the
County Council of Västerbotten.
references
[1] Sunnegårdh-Grönberg K, van Dijken JWV, Funegårdh U,
Lindberg A, Nilsson M. Selection of dental materials and
longevity of replaced restorations in Public Dental Health
clinics in northern Sweden. J Dent 2009;37:673–8.
[2] van Dijken JWV. All-ceramic restoratives: classification and
clinical evaluations. Compend Contin Educ Dent
1999;20:1115–34.
[3] Manhart J, Chen H, Hamm G, Hickel R. Review of the clinical
survival of direct and indirect restorations in posterior teeth
of the permanent dentition. Oper Dent 2004;29:
481–508.
[4] van Dijken JWV, Höglund-Åberg C, Olofsson AL. Fired
ceramic inlays. A six year follow up. J Dent 1998;26:219–25.
[5] van Dijken JWV, Örmin A, Olofsson AL. Pressed ceramic
inlays (IPS Empress) luted with a resin-modified glass
ionomer cement and a chemical-cured resin composite. J
Prosthet Dent 1999;82:529–35.
[6] Martin JA, Bader JD. Five-year treatment outcomesfor teeth
with large amalgams and crowns. Oper Dent 1997;22:72–8.
[7] Smales RJ, Hawthorne WS. Long-term survival of extensive
amalgam and posterior crowns. J Dent 1997;25:225–7.
[8] Plasmans PJJM, Creugers NHJ, Mulder J. Long-term survival of
extensive amalgam restorations. J Dent Res 1998;77:453–60.
[9] van Dijken JWV, Hasselrot L, Örmin A, Olofsson A-L.
Durability of extensive dentin–enamel-bonded ceramic
coverages (IPS Empress). A 5-year follow-up. Eur J Oral Sci
2001;109:222–9.
[10] Burke FJT. The effect of variations in bonding procedure on
fracture resistance of dentin-bonded all-ceramic crowns.
Quintessence Int 1995;26:293–300.
[11] Sjögren G, Molin M, van Dijken JWV. A 10-year prospective
evaluation of CAD/CAM-manufactured (Cerec) ceramic
inlays cemented with a chemically cured or a dual-cured
resin composite. Int J Prosthodont 2004;17:241–6.
[12] Felden A, Schmaltz G, Federlin M, Hiller K-A. Retrospective
clinical investigation and survival analysis on ceramic inlays
and partial ceramc crowns. Results up to 7 years. Clin Oral
Investig 1998;2:161–7.
[13] Smales RJ, Etemadi S. Survival of ceramic onlays placed with
and without metal reinforcement. J Prosthet Dent
2004;91:548–53.
[14] Frankenberger R, Taschner M, Garcia-Godoy F, Petschfelt A,
Krämer N. Leucite-reinforced glass ceramic inlays and
onlays after 12 years. J Adhes Dent 2008;5:
393–8.
[15] Pjetursson BE, Sailer I, Zwahlen M, Hämmerle CHF. A
systematic review of the survival and complication rates of
all-ceramic and metal-ceramic reconstructions after an
observation period of at least 3 years. Part I: Single crowns.
Clin Oral Implants Res 2007;18(Suppl. 3):
73–85.
[16] Borgmeijer PJ, Kreulen CM, Amerongen WE, Akerboom HBM,
Gruythuysen RMJ. The prevalence of postoperative
sensitivity in teeth restored with class II composite resin
restorations. J Dent Child 1991;58:378–83.
[17] Siegel S. Nonparametric statistics. New York: McGraw-Hill
Book Company, Inc.; 1956. pp. 166–172.
[18] Walton TR. An up to 15-year longitudinal study of 515
metal-ceramic FPDs: part 1. Outcome. Int J Prosthodont
2002;15:439–45.
[19] Creugers NH, Kayser AF, van ı̌Hof MA. A meta-analysis of
durability data on conventional fixed bridges. Community
Dent Oral Epidemiol 1998;22:448–52.
[20] Van Nieuwenhuysen J-P, Dı̌Hoore W, Carvalho J, Qvist V.
Long-term evaluation of extensive restorations in
permanent teeth. J Dent 2003;31:395–405.
[21] Malament KA, Socransky SS. Survival of Dicor-glass-ceramic
dental restorations over 14 years: Part 1. Survival of Dicor
complete coverage restorations and effect of internal
surface acid etching, tooth position, gender, and age. J
Prosthet Dent 1999;81:23–32.
[22] Reiss B, Walther W. Clinical long-term results and 10-year
Kaplan–Meier analysis of Cerec restorations. Int J
Computerized Dent 2000;3:9–23.
[23] Schulte AG, Vöckler A, Reinhardt R. Longevity of ceramic
inlays and onlays luted with a solely light-curing composite
resin. J Dent 2005;33:433–42.
[24] Fradeani M, Redemagni M. An 11-year clinical evaluation of
leucite-reinforced glass-ceramic crowns: a retrospective
study. Quintessence Int 2002;33:503–10.
[25] Felden A, Schmalz G, Hiller KA. Retrospective clinical study
and survival analysis on partial ceramic crowns; results up
to 7 years. Clin Oral Investig 2000;4:199–205.
[26] Marquardt P, Strub JR. Survival rates of IPS Empress 2
all-ceramic crowns and fixed partial dentures: results of a
5-year prospective clinical study. Quintessence Int
2006;37:253–9.
[27] Guess PC, Strub JR, Steinhart N, Wolkewitz M, Stappert CFJ.
All-ceramic partial coverage restorations-Midterm results of
a 5-year prospective clinical splitmouth study. J Dent
2009;37:627–37.
[28] Hayashi M, Wilson NHF, Yeung CA, Worthington HV.
Systematic review of ceramic inlays. Clin Oral Investig
2003;7:8–19.
[29] Krämer N, Taschner M, Lohbauer U, Petschfelt A,
Frankenberger R. Totally bonded ceramic inlays and onlays
after eight years. J Adhes Dent 2008;10:307–14.
[30] Naeselius K, Arnelund CF, Molin MK. Clinical evaluation of
all-ceramic onlays: a 4-year retrospective study. Int J
Prosthodont 2005;21:40–4.
[31] Lange RT, Pfeiffer P. Clinical evaluation of ceramic inlays
compared to composite restorations. Oper Dent
2009;34:263–72.
[32] Galiatsatos AA, Bergou D. Six-year evaluation of ceramic
inlays and onlays. Quintessence Int 2008;39:407–12.
[33] Lehner C, Studer S, Brodbeck U, Schärer P. Short-term result
of IPS-Empress full-porcelain crowns. J Prosthodontics
1997;6:20–30.
[34] Liebenberg WH. Use of resin-bonded partial coverage
ceramic restorations to treat incomplete fractures in
 d e n t a l m a t e r i a l s 2 6 ( 2 0 1 0 ) 929–939
posterior teeth: a clinical report. Quintessence Int
1996;27:739–47.
[35] Friedl K-H, Hiller K-H, Schmalz G. Placement and
replacement of amalgam restorations in Germany. Oper
Dent 1994;19:228–32.
[36] van Dijken JWV, Sunnegårdh-Grönberg K, Lindberg A.
Clinical long term retention of etch-and-rinse and self-etch
adhesive systems in non-carious cervical lesions. A 13 years
evaluation. Dent Mater 2007;23:1101–7.
[37] van Dijken JWV, Pallesen U. Long term dentin retention of
etch-and-rinse and self-etch adhesives and a resin modified
glass ionomer cement in non-carious cervical lesions. Dent
Mater 2008;24:915–22.
[38] van Dijken JWV. Clinical evaluation of three different
adhesive systems in Class V non-carious lesions. Dental
Mater 2000;16:285–91.
939
[39] Goodacre CJ, Bernal G, Rungcharassaeng K, Kan JYK. Clinical
complications in fixed prosthodontics. J Prosthet Dent
2003;90:31–41.
[40] Schulz Johansson A, Arvidsson K. A retrospective study of
Mirage ceramic inlays over 9 years. Int J Prosthodont
2003;16:510–4.
[41] De Backer H, Van Maele G, De Moor N, Van der Berghe L, De
Boever J. A 20-year retrospective survival study of fixed
partial dentures. Int J Prosthodont 2006;19:
143–53.
[42] Stoll R, Cappel I, Jablonski-Momeni A, Pieper K, Stachniss V.
Survival of inlays and partial crowns made of IPS Empress
after a 10-year observation period and in relation to various
treatment parameters. Oper Dent 2007;32:
556–63.