journal of dentistry 40 (2012) 443–452

Available online at www.sciencedirect.com

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

Review

Adhesives for the restoration of non-carious

cervical lesions: A systematic review

Brian Chee a,*, Luke J. Rickman b, Julian D. Satterthwaite c

a
Clinical Teaching Fellow in Fixed and Removable Prosthodontics, School of Dentistry, University of Manchester, Higher Cambridge Street,
Manchester M15 6FH, United Kingdom
b
General Dental Practitioner, Antley Villa Dental Practice, 432 Blackburn Road, Accrington, Lancashire BB5 0DE, United Kingdom
c
Restorative Dentistry, School of Dentistry, University of Manchester, Higher Cambridge Street, Manchester M15 6FH, United Kingdom

article info abstract

Article history: Objectives: To establish whether simplified adhesives (self-etch) are as clinically effective as
Received 1 November 2011 conventional adhesives (etch-and-rinse) with multiple application steps for treatment of
Received in revised form non-carious cervical lesions (NCCLs).
4 February 2012 Null hypothesis: there is no difference in the clinical effectiveness of the four different
Accepted 10 February 2012 bonding strategies: Three-step etch-and-rinse; Two-step etch-and-rinse; Two-step self-
etch; One-step self-etch for treatment of NCCLs.
Sources: Electronic databases were searched including: Cochrane Oral Health Group Trials
Keywords: Register, the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE and
Adhesives EMBASE. In addition, studies were identified by handsearching of selected journals.
Systematic review Study selection: Randomised controlled trials (RCTs) comparing at least two adhesives in
Non-carious cervical lesions non-carious cervical lesions (NCCLs), with at least 18 months follow-up were selected. The
Clinical effectiveness primary outcome was loss of retention/restoration loss, with marginal adaptation and
marginal discolouration as secondary outcomes. Criteria for quality assessment included:
random sequence generation; allocation concealment; blinding of outcome assessment;
and information on withdrawals. Twenty six studies were identified that met the inclusion
criteria. In general, studies were not of sufficient quality to fully address the objectives of
this review.
Conclusion: There is not enough evidence to support one adhesive or bonding strategy over
another for treatment of NCCLs. Consequently, the null hypothesis of no difference cannot
be supported or rejected with the data currently available. There is a need for better
standardisation and reporting of randomised controlled trials investigating adhesive per-
formance.
Clinical significance: Studies with low overall risk of bias demonstrated good clinical perfor-
mance for adhesives with all four bonding strategies. However, included studies showed
wide variation between adhesives of the same category.
# 2012 Elsevier Ltd. All rights reserved.

* Corresponding author at: Colgate Australian Clinical Dental Research Centre, 2nd Floor, Adelaide Dental Hospital, Frome Road, Adelaide,
SA 5001, Australia. Tel.: +61 8 8303 3436.
E-mail address: brian.chee@adelaide.edu.au (B. Chee).
0300-5712/$ – see front matter # 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jdent.2012.02.007
444 journal of dentistry 40 (2012) 443–452
1. Introduction
Direct resin composite restorations are commonplace in
dentistry and can provide predictable tooth-coloured restora-
tions.1,2 However, their success is reliant upon adhesion to the
underlying tooth via adhesive systems, which in turn facilitate a
more conservative approach to cavity preparation, less reliant
on the traditional removal of tooth structure for mechanical
retention and resistance form. The mechanism of adhesion is
essentially micromechanical to etched enamel or via the
formation of a hybrid layer with dentine.3,4 This conventionally
involves etching the tooth with 30–40% phosphoric acid,
followed by the application of a primer and subsequently an
adhesive resin: the 3-step etch-and-rinse approach.
The efficacy of three-step etch-and-rinse systems is well
established.5,6 However, many clinicians perceive these three-
step systems to have high technique sensitivity. A desire to
reduce technique sensitivity and clinical time has led manu-
facturers to develop simplified systems. Two-step etch-and-
rinse systems combine the primer and adhesive into one bottle
but maintain a separate etching step to remove the smear layer
and demineralise the surface layer of enamel and dentine. Self-
etch systems penetrate through the smear layer and incorporate
it into the hybrid layer to varying degrees dependent upon their
acidity. They consist of either a self-etching primer accompanied
by an adhesive resin applied as a subsequent step, or a self-etch
adhesive which does not require a separate primer. However,
concerns have been raised regarding the bonding performance
and durability of some simplified systems.7
Although the role of in vitro data to predict clinical
performance is increasingly recognised,8,9 randomised con-
trolled trials (RCTs) remain the most rigorous study design for
assessing the clinical effectiveness of an intervention. The
majority of studies investigating the clinical effectiveness of
bonding systems use the longevity of restorations in non-
carious cervical lesions (NCCLs) as the outcome. These lesions
are considered advantageous in the assessment of adhesive
systems.10 Amongst other benefits, they offer no mechanical
retention form and are located mainly in dentine, facilitating
evaluation of the resin–dentine bond which is less stable than
the resin–enamel bond. Previous systematic reviews9,11,12 have
investigated the clinical effectiveness of adhesives, although
detailed methods of quality assessment and a transparent
search protocol are not always available.
This study aims to evaluate the clinical effectiveness of
contemporary resin-based adhesive systems by systematic
review of the literature from 1950 to 2011. The specific
objective is to compare the various bonding strategies in the
restoration of NCCLs. The primary outcome of restoration
retention and the secondary outcomes of marginal integrity
and discolouration are employed to establish whether self-
etch adhesives are as clinically effective as etch-and-rinse
systems. The null hypothesis for this study was that no
differences would be detected between each adhesive type.
2. Materials and methods
RCTs that evaluated the clinical effectiveness of adhesives in
NCCLs were included in this systematic review. Multiple
reports of the same study were combined to make one entry.
Included studies had a minimum follow-up period of
18 months and compared at least two adhesive systems.
The primary outcome was loss of retention. Secondary
outcomes were marginal integrity and marginal discoloura-
tion.
Studies were considered for inclusion if they evaluated
resin-based adhesive systems used in conjunction with direct
composite resin. Adhesives were categorised according to the
classification described by Van Meerbeek et al.13 This
systematic review only evaluated adhesive systems with the
following bonding strategies:
1. Three-step etch-and-rinse
2. Two-step etch-and-rinse
3. Two-step self-etch
4. One-step self-etch
Eligible studies included participants who were healthy
adults in any age group (male and female) with NCCLs in the
permanent dentition. All settings were included (academic
university department, dental hospital, primary care, private
practice).
Studies were excluded if:
 Carious cervical lesions were treated.
 Other types of cavity were treated other than class V.
 Lesions had undercuts or were prepared with mechanical
retention.
 The adhesive was not used according to the instructions of
the manufacturer.
 The follow-up period was less than 18 months.
 Studies had extremely low recall rates (<75% after
18 months, <60% after 3 years and <50% after 5 years or
more).
 There was no appropriate control group (either another
resin-based adhesive or resin modified glass ionomer
cement).
 A bonding strategy was used other than those outlined
above.
To identify trials to be included for this review, a search
strategy developed for MEDLINE (via OVID) was revised
appropriately for each database used (Table 1). The search
protocol incorporated the Cochrane Highly Sensitive Search
Strategy for identifying reports of randomised controlled
trials.14 No language restrictions were applied.
The following electronic databases were searched on 8th
August 2011:
 Cochrane Oral Health Group Trials Register
 Cochrane Central Register of Controlled Trials (CENTRAL)
via the Cochrane Library, to current issue
 MEDLINE via OVID (1950–August 2011)
 EMBASE via OVID (1980–August 2011)
The reference lists of other systematic reviews and
identified trials were searched for additional relevant studies.
Conference abstracts were not included unless a full report of
 journal of dentistry 40 (2012) 443–452 445

Table 1 – Search strategy for Medline (OVID). year of publication. Data were extracted using customised
Search strategy 8/8/11 extraction forms and the following data recorded for each
1. non-carious.mp. included study:
2. class V.mp.
3. cervical.mp.  Details of the study including year of publication and
4. or/1–3
author(s).
5. lesion$.mp.
6. restoration$.mp.
 Details of study methods including follow-up period,
7. composite.mp. sequence generation, allocation concealment and blinding.
8. cavity.mp.  Details of participants including age, gender, setting and the
9. cavities.mp. inclusion criteria of the trial.
10. or/5–9  Details of the adhesives and composite resins used,
11. dental bonding/
including restorative protocol (cavity preparation, enamel
12. dentine-bonding agents/
bevelling, methods of isolation, and methods of polymeri-
13. bonding agent$.mp.
14. dentine$ bonding agent$.mp. sation).
15. adhesive$.mp. or adhesives/  Details of the outcomes including loss of retention, marginal
16. composite resin$.mp. or Composite Resins/ integrity, marginal discolouration, secondary caries, num-
17. (generation and (fourth or 4th or fifth or 5th or sixth ber of drop-outs and statistical analysis
or 6th or seventh or 7th)).mp.
18. (step and (three or ‘3’ or two or ‘2’ or one or single or ‘1’)).mp.
The quality of each included study was assessed in
19. ((‘etch and rinse’) or ‘etch & rinse’ or ‘self etch’
duplicate by two independent reviewers (BC and JS). Any
or ‘self-etch’ or ‘one bottle’ or ‘one-bottle’
or ‘all in one’ or ‘all-in-one’).mp. disagreement was resolved by discussion and if necessary by
20. or/11–19 referral to a third reviewer (LR).
21. randomized controlled trial.pt. The key criteria for judging risk of bias included the
22. controlled clinical trials.pt. following:
23. randomized.ab.
24. randomised.ab.
 Random sequence generation
25. randomly.ab.
26. trial.ab.
 Allocation concealment
27. groups.ab.  Blinding of outcome assessment
28. ‘split-mouth’.mp.  Incomplete outcome data including excluded data and drop-
29. or/21–28 outs
30. exp animals/ not humans.sh.  Selective outcome reporting
31. 29 not 30
32. 4 and 10 and 20 and 29 and 31
The quality assessment was pilot tested using a sample of
study reports to ensure that the criteria were applied
eligible studies could be obtained. Manufacturers of adhesive consistently. For each aspect of the quality assessment, the
systems and selected study authors were contacted to identify risk of bias was scored following recommendations as
unpublished and ongoing trials. described in the Cochrane Handbook for Systematic reviews of
The following journals known to publish clinical trials on Interventions 5.0.2. The judgement for each entry involved
adhesives were identified: recording ‘yes’ indicating low risk of bias, ‘no’ indicating high
risk of bias, and ‘unclear’ indicating either lack of information
 Acta Odontologica Scandinavica or uncertainty over the potential for bias.
 British Dental Journal Studies were considered to be at low risk of bias if there was
 Dental Materials adequate sequence generation, blinded outcome assessment
 Journal of Adhesive Dentistry and adequate information on withdrawals. If one or more of
 Journal of the American Dental Association (JADA) these criteria were not met a study would be considered at
 Journal of Dental Research moderate or high risk of bias, respectively.
 Journal of Dentistry The extracted data were analysed using Revman 5 (Review
 Journal of Prosthetic Dentistry Manager ver. 5, The Cochrane Collaboration, Copenhagen,
 Operative Dentistry Denmark) and reported according to Cochrane Collaboration
criteria14 and PRISMA guidelines.15 Data from eligible studies
Journals not already handsearched by the Cochrane Oral were either dichotomous (loss of retention, occurrence of
Health Group handsearching program were handsearched by secondary caries) or ordinal (marginal adaptation and mar-
one reviewer. Electronic searches were screened by one ginal discolouration). Ordinal data (mostly modified USPHS
reviewer to identify eligible studies. Full text versions of the criteria)16 were dichotomised into either ‘clinically acceptable’
papers that appeared to meet the inclusion criteria were or ‘clinically unacceptable’, i.e. alpha and bravo scores were
retrieved for further assessment and data extraction. Full categorised as clinically acceptable, while charlie scores or
reports were also obtained when there was insufficient lower were considered clinically unacceptable.
information in the title and abstract to make a clear decision. Comparisons were made between adhesive systems and
To aid the management of a large number of studies, each between the four types of bonding strategy. The following data
eligible study was given a study ID, combining first author and syntheses were planned if available data allowed:
446 journal of dentistry 40 (2012) 443–452

Table 2 – Characteristics of excluded studies. from the report or retrieved by the author were not included in
Reason for exclusion Number of the data synthesis.
excluded papers Trials were assessed for clinical heterogeneity. If they were
similar (in terms of participants, interventions and outcomes
Randomised allocation not reported 11
Less than 18 months follow-up period 23 measured) formal assessment of heterogeneity was planned
Lack of adequate control 23 using chi-squared test and I2 statistic. Subgroup analysis was
Manufacturers instructions not followed 3 carried out to examine the effect of follow-up period and risk
Cavities other than NCCL used 9 of bias on the results.
Low recall rate 2
Adhesive(s) investigated were 6
not contemporary 3. Results
Total 77

One hundred and three studies were identified. Following

 For each trial risk ratios (RR) and their 95% confidence
intervals calculated for dichotomous data (primary and
secondary outcomes).
 Number of lesions needed to treat (NNT) to prevent loss of
retention in one class V restoration.
 If appropriate, meta-analysis was planned by pooling the
risk ratios of the individual trials using a random-effects
model in cases where there were three or more trials. If
fewer trials were included for a comparison, then the fixed
effects model would be used.
Following the data extraction and quality assessment
process, trial authors were contacted in order to obtain
further information of any unclear or missing data. Included
studies for which the necessary data could not be extracted
Identification
assessment, 77 of these were excluded for the reasons listed in
Table 2. In the 26 studies identified for inclusion, 13 were
designed as parallel group studies, 11 had a paired design and
two had quadrant allocation. An overview of the number of
study reports screened, assessed and included in the review
can be seen in Fig. 1.
The majority of included studies reported university dental
hospitals as the research setting. The mean age of participants
across studies was 52.9 (SD 6.0), although this did not take into
account four studies where median age or age ranges were
reported,17–20 and three studies where age was not
reported.6,21,22 All patients were adult and had NCCLs,
although Ritter et al.23 and Turkun24 both included two
lesions that had superficial areas of caries removed.
Details of adhesive systems used in the included trials can
be seen in Table 3. Allbond 3 and Allbond SE are of interest, as

Records identified through database Additional records identified

searching through other sources
719 3

Records after duplicates removed

721
Screening*

Records screened Records excluded

114 11
Eligibility

Full-text articles assessed for eligibility Full-text articles excluded,

with reasons
103 77

Studies included in qualitative

synthesis
26
Included

Studies included in quantitative

synthesis (meta-analysis)
0

* records remaining after removal of non-relevant articles

Fig. 1 – Flow diagram of study identification.15

Table 3 – Characteristics of included studies.
Study ID Study design Participants Restorations Recall Mean age Dentine Enamel Isolation Adhesive(s)
period prep bevel
(months)
Aw et al.33 Parallel group 57 171 18 51 N Y Retraction cord Scotch-Bond Multi-Purpose, Single
Bond, One Coat Bond
Brackett et al.17 Paired design 24 74 24 47 N N Retraction cord Single Bond
Brackett et al.18 Paired design 25 76 18 Median 52 N N Retraction cord Tyrian, One-Step

journal of dentistry 40 (2012) 443–452

Brackett et al.19 Paired design 14 80 24 Median 46 N N Retraction cord Clearfil SE Bond, Clearfil S3 Bond
Kim et al.34 Parallel group 39 150 24 50 Y Y Retraction cord Scotch-Bond Multi-Purpose,
Adper Prompt
Kubo et al.35 Quadrant 8 72 60 61.3 Y Y Retraction cord Clearfil Liner Bond II, Single Bond
Kubo et al.36 Quadrant 23 108 24 61.8 Y Y Retraction cord S3 Bond, G-Bond
Loguercio et al.27 Paired design 25 78 36 Range reported N N Rubber dam Single Bond, Adper Prompt
Loguercio et al.30 Paired design 33 66 24 Range reported N N Rubber dam All Bond SE (one-step), All Bond
SE (two-step)
Matis et al.26 Paired design 30 80 36 45 N Y Rubber dam FL-Bond, Scotch-Bond Multi-Purpose
McCoy et al.25 Parallel group 23 126 36 48 N N Rubber dam All-Bond 2, ART Bond, Prisma
Universal Bond 3
Merte et al.21 Parallel group 80 89 24 ? N N Rubber dam Prime & Bond 2.1
Neo et al.40 Parallel group 10 83 18 47 N N Retraction cord Prisma Universal Bond 3, Imperva Bond
Onal and Pamir22 Parallel group 30 130 24 ? Y N Cotton rolls, suction Scotch-Bond Multi-Purpose
37
Ozgunaltay and Onen Paired design 24 98 36 Range reported N Y Retraction cord Scotch-Bond Multi-Purpose
Peumans et al.6 Paired design 71 142 84 ? N Y Rubber dam Permaquick, Optibond FL
Reis and Loguercio29 Parallel group 84 84 36 Range reported N N Rubber dam Single Bond, One Step
Reis et al.31 Paired design 33 66 24 Range reported N N Rubber dam All-Bond 3 (2-step), All-Bond 3 (3-step)
Ritter et al.28 Parallel group 30 105 36 55 Y Y Rubber dam iBond, Gluma Solid Bond
Ritter et al.23 Parallel group 33 99 96 53 Y Y Retraction cord Optibond Solo, Prime & Bond 2.1
Santiago et al.32 Paired design 30 70 24 Range reported N N Rubber dam Excite
Turkun24 Paired design 32 98 24 46 Y N Retraction cord Clearfil SE Bond, Prime & Bond NT
Van Dijken38 Parallel group 60 148 36 57 Y N Retraction instrument EBS, One-Step
Van Dijken39 Parallel group 90 144 24 58 Y N Retraction instrument Clearfil Liner Bond II, One Coat
Bond, Prompt L-pop
Van Dijken44 Parallel group 72 119 96 60.1 N N Retraction instrument Clearfil SE Bond, PQ1
Van Landuyt et al.20 Parallel group 52 267 36 Range reported Y Y Retraction cord/matrix G-Bond, Optibond FL

447
448 journal of dentistry 40 (2012) 443–452

Table 4 – Marginal discolouration: clinically acceptable Table 5 – Marginal integrity: clinically acceptable re-
restorations (%) for adhesive type across studies. storations (%) for adhesive type across studies.
Reported percentages (%) Reported percentages (%)
3-Step etch-and-rinse 3-Step etch-and-rinse
Clearfil Liner Bond II 10035, 10039 Clearfil Liner Bond II 9139, 10035
Optibond FL 946, 10020 Optibond FL 1006, 10020
Permaquick 946, 976 Permaquick 1006, 1006
Scotch-Bond Multi-Purpose 8733, 90(a)26, 10022, 10034, 10037 Scotch-Bond Multi-Purpose 73(a)26, 9633, 10022, 10034, 10037
Imperva Bond 9240 Imperva Bond 9240
EBS 10038 EBS 9038
Allbond 3 10031 Allbond 3 10031
GLUMA Solid Bond 100(a)28 GLUMA Solid Bond 92(a)28
2-Step etch-and-rinse 2-Step etch-and-rinse
One Coat Bond 8633, 10039 One Coat Bond 8739, 8833
One-step 10018, 10029, 10038 One-Step 5138, 10018, 10029
Prime & Bond 2.1 94(a)23, 10021 Prime & Bond 2.1 9521, 100(a)23
Prime & Bond NT 10024 Prime & Bond NT 10024
Single Bond 83(a)27, 9533, 10017, 10035, 10029 Single Bond 83(a)27, 9533, 9735, 10017, 10029
Optibond Solo 91(a)23 Optibond Solo 91(a)23
Excite 10032 Excite 9732
2-Step self-etch 2-Step self-etch
Tyrian 10018 Tyrian 10018
Allbond SE 10030 Allbond SE 10030
Prisma Universal Bond 3 10040 Prisma Universal Bond 3 10040
Clearfil SE Bond 10019, 10024, 10044 Clearfil SE Bond 7644, 10019, 10024
1-Step self-etch 1-Step self-etch
G-Bond 10020, 10036 G-Bond 10036, 10020
S3 Bond 10019, 10036 S3 Bond 10019, 10036
Adper Prompt 53(a)27, 10034 Adper Prompt 67(a)27, 10034
Prompt L-Pop 10039 Prompt-L-Pop 7939
iBond 35(a)28, 69(a)28 iBond 70(a)28, 81(a)28
(a) = percentage alpha scores reported. (a) = percentage alpha scores reported.

they can be used with or without the application of a separate
bonding resin layer, although the manufacturer recommends
its application for improved performance. The majority of
studies reported cleaning of NCCLs with pumice and water
prior to bonding. Variation in clinical technique was noted,
with 10 studies isolating with rubber dam6,21,25–32 and the
remaining studies isolating with a combination of cotton rolls,
saliva ejector and retraction cord or other retraction instru-
ments. Ten of the included studies prepared the NCCLs with
an enamel bevel,6,20,23,26,28,33–37 and ten of the studies prepared
dentine by bur roughening.20,22–24,28,34–36,38,39 Some studies
investigated materials in addition to resin-based adhesives,
including resin-modified glass ionomer cement
(RMGIC)17,22,37,38,40 and poly-acid modified composite resins
(compomers).22,25 Data for these materials were not used in
this review, although resin-modified glass ionomers were
considered an appropriate control group.
Most studies used modified USPHS criteria with two studies
using very similar Vanherle criteria.6,20 The duration of the
studies varied from 18 months to 8 years. Data for the primary
outcome were presented as numbers of restorations lost, as
well as retention rates or cumulative failure rates (%).
Reporting of the secondary outcomes varied, but data could
be dichotomised into ‘clinically acceptable’ or ‘clinically
unacceptable’ except for five studies23,25–28 (Tables 4 and 5).
Cohen’s kappa coefficient was calculated for measuring
agreement between reviewers (BC, JS) and was considered
‘excellent’ (0.88). Overall, only 23% of included studies clearly
described adequate sequence generation. Allocation conceal-
ment was not reported in any of the included studies and 58%
reported blind evaluation. In 62% of the included studies,
information was provided on participants/restorations not
available at follow-up, with the reasons given. Eighty one
percent of studies were free of selective reporting, but only
seven out of 26 studies (27%) were free of other bias. The most
common reasons for this were either a lack of operator/
assessor calibration or failure to account for clustering within
participants. Summaries of the overall risk of bias are
presented graphically in Figs. 2 and 3. For the primary
outcome, only five studies were considered to have low risk
of bias for all key domains.6,20,29–31
The worst clinical performance reported in terms of
marginal integrity was found for One-Step with 51% of
restorations considered clinically acceptable at 36 months.38
Fig. 2 – Methodological quality graph: judgements about
each methodological quality item presented as
percentages across all included studies.
 journal of dentistry 40 (2012) 443–452
Fig. 3 – Methodological quality summary: judgements
about each methodological quality item for each study.
449
However, Bracket et al. as well as Reis et al. found 100% clinical
acceptability for this adhesive at 18 and 36 months, respec-
tively.18,29 One Coat Bond had the poorest reported clinical
performance in terms of marginal discolouration, with 86% of
restorations clinically acceptable at 18 months, although there
was no significant difference between adhesives ( p > 0.05).33
The percentages of clinically acceptable restorations in terms
of marginal discolouration and marginal integrity are sum-
marised for each adhesive system in Tables 4 and 5.
4. Discussion
4.1. Summary of main results
The studies included in this review evaluated 26 resin-based
adhesives and recruited 1032 participants. There was variation
in study methodology and quality, although several of the
designs were based on acceptance programme guidelines of
The American Dental Association, which have since been
discontinued.41 The primary outcome was often presented as
retention rates or cumulative failure rates (%), calculated
according to the ADA recommended equation:
previous failures þ new failures
Failure percentage ¼ 100% 
previous failures þ currently recalled restorations
Only five studies were considered to have a low overall risk
of bias6,20,29–31 and the other studies were generally not of
sufficient quality to address the objectives of this review.
Frequently, data from studies were either reported inade-
quately or were inappropriate for use in meta-analyses.
Hence, this review was limited to the qualitative description
of studies. It should be noted that several excluded studies had
robust designs for their specific research question but did not
have an adequate control for the objectives of this systematic
review. For example, studies that investigated a single
adhesive under different conditions, or investigated a single
adhesive in conjunction with different resin composites had
no effective control regarding the adhesive system, and were
therefore excluded.
The planned analyses such as risk ratios, number needed to
treat and statistical heterogeneity (chi-squared test and I2
statistic) were not possible with the available data. A large
proportion of the raw data required for the proposed analyses
was not reported or could not be extracted from reported
percentages. Trial authors were contacted to request the
original data, although there was a poor response. A number of
studies used the restoration as the unit of analysis, rather than
the patient and many studies had multiple restorations per
patient. Although the majority of studies controlled for the
effects of clustering by limiting the number of restorations per
patient (typically less than 3), few studies accounted for the
clustering in the statistical analysis and the fact that failure
events were not independent. It was not considered appropri-
ate to use data that did not take into account this clustering
within patients, as these ‘units of analysis errors’ can produce
over-precise results and p-values that are too small.14
Although it has been argued that including multiple restora-
tions per patient is a valid way of increasing sample size
provided that appropriate analyses are performed 8 there
450 journal of dentistry 40 (2012) 443–452
should be caution if using this data in meta-analyses, as
uncorrected unit of analysis errors would cause the data to
receive too much weighting.
4.2. Overall completeness and applicability of the evidence
Risk of bias was mostly assessed at the outcome level, since
the criteria have different relevance to primary and secondary
outcomes, i.e. the blinding of outcome evaluation is less
important for restoration retention, as this outcome is very
objective. In contrast, inadequate blinding of outcome
assessment has greater risk of introducing bias in the more
subjective outcomes (marginal integrity, marginal discoloura-
tion). It should be noted that a potential source of bias in a
systematic review investigating restoration longevity is that
the USPHS criteria may not be consistently applied between
studies.42
The most objective outcome to assess the clinical effec-
tiveness of adhesive systems is retention of the class V
composite restoration, since bond failure will cause loss of the
restoration (in the absence of mechanical retention features).
In this study, the inclusion of marginal discolouration and
marginal integrity as outcomes measures is also essential,
since the breakdown of restoration margins is a common
reason for the replacement and repair of adhesive restora-
tions.13
Blinding of the clinician is unfeasible in most trials that test
adhesives, as the specific clinical protocol needs to be known
for each adhesive and is difficult to conceal. Blinding of the
participant was explicitly described in only seven stud-
ies.20,22,27,29–31,33 However, lack of patient blinding is unlikely
to introduce bias as there were no patient-centred outcomes
and their behaviour is unlikely to change as a result of
knowledge of the intervention groups. Sample size calcula-
tions were described by only four of the included stud-
ies.23,27,30,31 Although trial authors and manufacturers were
contacted to identify ongoing and unpublished trials, the
response was poor and none of the included studies were
identified by this method.
4.3. Agreement and disagreement with other research
The conclusions of other systematic reviews that investi-
gate the clinical performance of adhesives could not be
supported by this review. Peumans et al.11 found less
favourable clinical performance in the self-etch bonding
strategies compared with an etch-and-rinse protocol.
Superior performance of etch-and-rinse systems was also
reported by Heintze et al.9 in a systematic review of clinical
trials from 1994 to 2008. Krithikadatta12 published a
systematic review addressing the same research question
including studies from 2004 to 2010 and concluded that the
clinical performance of different categories of bonding
system were comparable.
Not all the reviews above present full details regarding their
methodology and review/assessment process; however, the
different conclusions reached in these studies may be
explained by the rigorous selection criteria and robust quality
assessment applied in the present review, with only high
quality evidence being considered.
4.4. Implication for research
This review has highlighted the need for better design,
implementation and reporting of RCTs. To facilitate meta-
analysis studies should follow the CONSORT guidelines43 and
reports should include failure as the numbers of restorations
lost. New adhesives should continue to be compared against
an established three-step etch-and-rinse adhesive, since they
are supported by the most long-term clinical and laboratory-
based evidence.8
4.5. Implications for practice
The included studies with low overall risks of bias6,20,29–31
reported favourable medium-term (2–3 years) clinical perfor-
mance for all four bonding strategies in terms of restoration
retention. Optibond FL also showed excellent long-term
performance with retention rates of 96% and 92%, respective-
ly, at 3 and 7 years.6,20 In these studies, five currently available
adhesives (Allbond 3, Allbond SE, G-Bond, Optibond FL,
Permaquick) exceeded the performance needed for accep-
tance by the now discontinued ADA standards.41 The included
studies found wide variation between the clinical perfor-
mances of adhesives with the same bonding strategy.
5. Conclusion
There was insufficient evidence to make firm recommenda-
tions for the use of one adhesive system or bonding strategy
over another. The proportion of information obtained from
studies with an unclear or high risk of bias was high. The null
hypothesis of no difference could not be supported or rejected
with the data currently available, using the robust analysis
planned. Studies with low overall risk of bias found good
clinical performance for adhesives with three-step etch-and-
rinse, two-step etch-and-rinse, two-step self-etch and one-
step self-etch bonding strategies.6,20,29–31 There is a need for
future research on the clinical effectiveness of adhesives. In
particular, better standardisation and reporting of randomised
controlled trials will allow more meaningful comparisons to
be made between adhesives.
references
1. Manhart J, Chen HY, Hamm G, Hickel R. Review of the
clinical survival of direct and indirect restorations in the
permanent dentition. Operative Dentistry 2004;29:481–508.
2. Opdam NJ, Bronkhurst EM, Loomans BA, Huysmans MC. 12-
year survival of composite vs. amalgam restorations. Journal
of Dental Research 2010;89:1063–7.
3. Buonocore. A simple method of increasing the adhesion of
acrylic filling materials to enamel surfaces. Journal of Dental
Research 1955;34:849–53.
4. Nakabayashi N, Kojima K, Masuhara E. The promotion of
adhesion by the infiltration of monomers into tooth
substrates. Journal of Biomedical Materials Research 1982;16:
265–73.
5. De Munck J, Van Landuyt K, Peumans M, Poitevin A,
Lambrechts P, Braem M, et al. A critical review of the
 journal of dentistry 40 (2012) 443–452
durability of adhesion to tooth tissue: methods and results.
Journal of Dental Research 2005;84:118–32.
6. Peumans M, De Munck J, Van Landuyt KL, Kanumilli P,
Yoshida Y, Inoue S, et al. Restoring cervical lesions with
flexible composites. Dental Materials 2007;23:749–54.
7. Van Landuyt K, Mine A, De Munck J, Jaecques S, Peumans M,
Lambrechts P, et al. Are one-step adhesives easier to use and
better performing? Multifactorial assessment of
contemporary one-step self-etching adhesives. Journal of
Adhesive Dentistry 2009;11:175–90.
8. Van Meerbeek B, Peumans M, Poitevin A, Mine A, Van
Ende A, Neves A, et al. Relationship between bond-
strength tests and clinical outcomes. Dental Materials
2010;26:e100–21.
9. Heintze SD, Thunpithayakul C, Armstrong SR, Rousson V.
Correlation between microtensile bond strength data and
clinical outcome of Class V restorations. Dental Materials
2011;27:114–25.
10. Van Meerbeek B, Perdigao J, Lambrechts P, Vanherle G. The
clinical performance of adhesives. Journal of Dentistry
1998;26:1–20.
11. Peumans M, Kanumilli P, De Munck J, Van Landuyt K,
Lambrechts P, Van Meerbeek B. Clinical effectiveness of
contemporary adhesives: a systematic review of current
clinical trials. Dental Materials 2005;21:864–81.
12. Krithikadatta J. Clinical effectiveness of contemporary
dentin bonding agents. Journal of Conservative Dentistry
2010;13:173–83.
13. Van Meerbeek B, De Munck J, Yoshida Y, Inoue S, Vargas M,
Vijay P, et al. Buonocore memorial lecture – adhesion to
enamel and dentin: current status and future challenges.
Operative Dentistry 2003;28:215–35.
14. Higgins J, Green S. Cochrane Handbook for Systematic
Reviews of Interventions 5.0.2 [updated September 2009].
2009. Available from: www.cochrane-handbook.org [cited
March 2010].
15. Moher D, Liberati A, Tetzlaff J, Altman DG, Group TP.
Preferred reporting items for systematic reviews and meta-
analyses: the PRISMA statement. PloS Medicine
2009;6:e1000097.
16. Cvar JF, Ryge G. Reprint of criteria for the clinical evaluation
of dental restorative materials. Clinical Oral Investigations
2005;9:215–32.
17. Brackett WW, Dib A, Brackett MG, Reyes AA, Estrada BE.
Two-year clinical performance of Class V resin-modified
glass-lonomer and resin composite restorations. Operative
Dentistry 2003;28:477–81.
18. Brackett WW, Brackett MG, Dib A, Franco G, Estudillo H.
Eighteen-month clinical performance of a self-etching
primer in unprepared class V resin restorations. Operative
Dentistry 2005;30:424–9.
19. Brackett MG, Dib A, Franco G, Estrada BE, Brackett WW.
Two-year clinical performance of Clearfil SE and Clearfil S3
in restoration of unabraded non-carious class V lesions.
Operative Dentistry 2010;35:273–8.
20. Van Landuyt K, Peumans M, De Munck J, Cardoso MV, Ermis
B, Van Meerbeek B. Three-year clinical performance of a
HEMA-free one-step self-etch adhesive in non-carious
cervical lesions. European Journal of Oral Sciences
2011;119:511–6.
21. Merte K, Frohlich M, Hafer M, Hirsch E, Schneider H,
Winkler M. Two-year clinical performance of two primer
adhesives on class V restorations. Journal of Biomedical
Materials Research 2000;53:93–9.
22. Onal B, Pamir T. The two-year clinical performance of
esthetic restorative materials in noncarious cervical
451
lesions. Journal of the American Dental Association
2005;136:1547–55.
23. Ritter AV, Swift Jr EJ, Heymann HO, Sturdevant JR, Wilder Jr
AD. An eight-year clinical evaluation of filled and unfilled
one-bottle dental adhesives. Journal of the American Dental
Association 2009;140:28–37. quiz 111–2.
24. Turkun SL. Clinical evaluation of a self-etching and a one-
bottle adhesive system at two years. Journal of Dentistry
2003;31:527–34.
25. McCoy RB, Anderson MH, Lepe X, Johnson GH. Clinical
success of class V composite resin restorations without
mechanical retention. Journal of the American Dental
Association 1998;129:593–9.
26. Matis BA, Cochran MJ, Carlson TJ, Guba C, Eckert GJ. A
three-year clinical evaluation of two dentin bonding
agents. Journal of the American Dental Association
2004;135:451–7.
27. Loguercio AD, Bittencourt DD, Baratieri LN, Reis A. A 36-
month evaluation of self-etch and etch-and-rinse adhesives
in noncarious cervical lesions. Journal of the American Dental
Association 2007;138:507–14. quiz 35–7.
28. Ritter AV, Heymann HO, Swift Jr EJ, Sturdevant JR, Wilder Jr
AD. Clinical evaluation of an all-in-one adhesive in non-
carious cervical lesions with different degrees of dentin
sclerosis. Operative Dentistry 2008;33:370–8.
29. Reis A, Loguercio AD. A 36-month clinical evaluation of
ethanol/water and acetone-based etch-and-rinse adhesives
in non-carious cervical lesions. Operative Dentistry
2009;34:384–91.
30. Loguercio AD, Manica D, Ferneda F, Zander-Grande C,
Amaral R, Stanislawczuk R, et al. A randomized clinical
evaluation of a one- and two-step self-etch adhesive over 24
months. Operative Dentistry 2010;35:265–72.
31. Reis A, Manica D, Ferneda F, Amaral R, Stanislawczuk R,
Manso A, et al. A 24-month randomized clinical trial of a
two- and three-step etch-and-rinse technique. American
Journal of Dentistry 2010;23:231–6.
32. Santiago SL, Passos VF, Vieira AHM, Navarro MFdL. Lauris
JRP, Franco EB. Two-year clinical evaluation of resinous
restorative systems in non-carious cervical lesions. Brazilian
Dental Journal 2010;21:229–34.
33. Aw TC, Lepe X, Johnson GH, Mancl LA. A three-year clinical
evaluation of two-bottle versus one-bottle dentin
adhesives. Journal of the American Dental Association
2005;136:311–22.
34. Kim S-Y, Lee K-W, Seong S-R, Lee M-A, Lee I-B, Son H-H,
et al. Two-year clinical effectiveness of adhesives and
retention form on resin composite restorations of
non-carious cervical lesions. Operative Dentistry
2009;34:507–15.
35. Kubo S, Kawasaki K, Yokota H, Hayashi Y. Five-year clinical
evaluation of two adhesive systems in non-carious cervical
lesions. Journal of Dentistry 2006;34:97–105.
36. Kubo S, Yokota H, Yokota H, Hayashi Y. Two-year clinical
evaluation of one-step self-etch systems in non-carious
cervical lesions. Journal of Dentistry 2009;37:149–55.
37. Ozgunaltay G, Onen A. Three-year clinical evaluation of a
resin modified glass-ionomer cement and a composite resin
in non-carious class V lesions. Journal of Oral Rehabilitation
2002;29:1037–41.
38. Van Dijken JW. Clinical evaluation of three adhesive
systems in class V non-carious lesions. Dental Materials
2000;16:285–91.
39. Van Dijken JWV. Durability of three simplified adhesive
systems in Class V non-carious cervical dentin lesions.
American Journal of Dentistry 2004;17:27–32.
452 journal of dentistry 40 (2012) 443–452

40. Neo J, Chew CL, Yap A, Sidhu S. Clinical evaluation of tooth-
colored materials in cervical lesions. American Journal of
Dentistry 1996;9:15–8.
41. ADA. Acceptance program guidelines for dentin and enamel
adhesive materials. Chicago: American Dental Association;
2001
. two-step self-etching adhesive and a heavily filled two-step
42. Chadwick B, Treasure E, Dummer P, Dunstan F, Gilmour A,
Jones R, et al. Challenges with studies investigating
longevity of dental restorations – a critique of a systematic
review. Journal of Dentistry 2001;29:155–61.
43. Moher D, Schulz KF, Altman DG. The CONSORT statement:
revised recommendations for improving the quality of
reports of parallel-group randomized trials. Annals of Internal
Medicine 2001;134:657–62.
44. Van Dijken JWV. A prospective 8-year evaluation of a mild
etch-and-rinse system in non-carious cervical lesions.
Dental Materials 2010;26:940–6.