A Systematic Review of Screw- versus Cement-Retained

Implant-Supported Fixed Restorations

Sami Sherif, BDS, DMSc,1,2 Harlyn K. Susarla, DMD, MPH,1 Theodoros Kapos, DMD, MMSc,1,2
Deborah Munoz, DMD, MMSc,1,3 Brian M. Chang, DDS, FACP,4 & Robert F. Wright, DDS, FACP5
1
Harvard School of Dental Medicine, Boston, MA
2
Private Practice, London, United Kingdom
3
Private Practice, Boston, MA
4
Department of Maxillofacial Prosthodontics, Cleveland Clinic, Cleveland, OH
5
Department of Prosthodontics, University of North Carolina School of Dentistry, Chapel Hill, NC

Keywords
Systematic review; implant failure; crown
failure; screw loosening; decementation;
porcelain fracture; screw-retained
restorations; cement-retained restorations.
Correspondence
Robert F. Wright, UNC School of Dentistry,
335 Brauer Hall, Campus Box 7450, Chapel
Hill, NC 27599-7450. E-mail:
robert_wright@dentistry.unc.edu
The authors deny any conflicts of interest.
Accepted August 21, 2013
doi: 10.1111/jopr.12128
Abstract
Purpose: To systematically evaluate the survival and success of screw- versus
cement-retained implant crowns.
Materials and Methods: The authors performed an electronic search of nine
databases using identical MeSH phrases. Systematic evaluation and data extraction
of the articles from 1966 through 2007 were completed by three reviewers and two
clinical academicians. The major outcome variable was implant or crown loss, and the
minor outcome variables were screw loosening, decementation, and porcelain fracture.
Random effects Poisson models were used to analyze the failure and complication
rates.
Results: The initial search produced 26,582 articles. Of these, 577 titles and sub-
sequently 295 abstracts were available for evaluation, with 81 full texts meeting the
criteria for review. Data were extracted from 23 level one and two research studies.
Fleiss’ kappa interevaluator agreement ranged from almost perfect to moderate. Ma-
jor failures included 0.71 screw-retained and 0.87 cement-retained failures per 100
years. Minor failures included 3.66 screw loosenings, 2.54 decementations, and 0.46
porcelain fractures per 100 years.
Conclusion: There is no significant difference between cement- and screw-retained
restorations for major and minor outcomes with regard to implant survival or crown
loss. This is important data, as clinicians use both methods of restoration, and neither
is a form of inferior care.

The early modern era of endosseous implant therapy was dom-
inated by the screw-retained restoration. Such rehabilitations,
which were initially intended for the edentulous patient, were
mostly of a full-arch nature. The initial “ad modem Branemark”
protocol called for an edentulous patient to be treated with four
to six 3.75 mm external hex implants placed in the anterior
mandible. The anterior mandible was selected for several rea-
sons. As the lower anterior teeth are usually the last to be lost,
a greater volume of bone exists in this area. This increased vol-
ume allows for the use of longer implants, ultimately providing
more bicortical stabilization. The intraforaminal placement of
the implants in the anterior mandible also avoids the inferior
alveolar nerve in addition to reducing the effects of mandibular
flexion, which occurs mostly in the posterior mandible up to a
magnitude of 800 µm upon opening.1 The implants were cov-
ered for 4 to 6 months, and subsequently restored with a screw-
retained gold bar overlaid with pink acrylic and denture teeth.
Screw-retained crowns were chosen because they arguably of-
fer more reliable retrieval, have a decreased space requirement,
and result in healthier soft tissues, as no cement cleanup is
necessary.2-4 The use of acrylic denture teeth not only simpli-
fies maintenance of the prosthesis, but is also thought to provide
a dampening force on the implants from occlusal trauma.
As the scope of implant therapy was increased to include
treating the partially edentulous patient, the cement-retained
restoration gradually became more popular. The 1988 intro-
duction of the UCLA custom abutment, which permitted the
retention of a prosthesis directly on the implant without the
use of a transmucosal abutment, allowed for smaller interoc-
clusal space requirements.5 Telescopic crowns were then fab-
ricated on these abutments. Subsequently, the introduction of
a screw-retained abutment with a cemented restoration, Cera
One (Nobel Biocare, Yorba Linda, CA), enhanced the suc-
cess of implant therapy.6 Cement-retained crowns offered the

Journal of Prosthodontics 23 (2014) 1–9 

C 2013 by the American College of Prosthodontists 1
Screw- versus Cement-Retained Implant-Supported Fixed Restorations
Figure 1 Publications retained at title, abstract, and full-text review stages.
clinician improved occlusal accuracy, enhanced esthetics, in-
creased chances of achieving a passive fit, and decreased
instances of retention loss. They were more akin to conven-
tional fixed prosthodontics and were less costly to fabricate.7
Though there is an abundance of retrospective and prospec-
tive studies evaluating placement of screw- and cement-retained
restorations, there is a dearth of systematic assessments of their
outcomes. As such, the purpose of this review is to system-
atically evaluate the outcomes of screw- and cement-retained
implant restorations.
Materials and methods
Systematic search design and article selection
The search strategy was designed to identify level 1 and level
2 evidence of the outcomes of screw- and cement-retained
restorations in healthy patients with partial edentulism treated
with fixed prosthodontic implant therapy. Interventions were
broadly classified into two groups: screw-retained or cement-
retained restorations. To be included, eligible studies must have
had a follow-up period of at least 12 months.
The outcomes of interest were classified as major and minor
outcomes. Major outcomes included those factors leading to
restoration failure (i.e., failure of the prosthesis, thus requir-
ing replacement). These included abutment fracture, esthetic
failure, severe prosthesis fracture, and implant failure. As a
2 Journal of Prosthodontics 23 (2014) 1–9 
Sherif et al
function of time, these outcomes measures were represented as
exposure time in months. Failures of implant fixtures preloaded
with definitive restorations were excluded.
Minor outcome factors were classified as those requiring
clinician intervention that immediately threatened survival of
the restorations. Included in this category were screw loosening,
decementation and subsequent total loss of retention, porcelain
fractures that did not necessitate replacement of the prosthesis,
bone loss per month, strain, and marginal gap discrepancies.
Search strategy
The search strategy (Fig 1) began with an electronic search of
publications from 1966 to 2007. This search was performed
using the following electronic databases: MEDLINE (1966
to December 2007), EMBASE (1980 to December 2007), the
Cochrane Oral Health Group Trials Register, and the Cochrane
Central Register of Controlled Trials (CENTRAL). The search
included only English language articles published in peer-
reviewed journals. The keywords used for the search were
combinations of the following:
1. “Dental implant”
2. “Screw-retained crown OR prosthesis”
3. “Cement-retained crown OR prosthesis”
4. “implant crown esthetics”
5. “implant crown satisfaction”
C 2013 by the American College of Prosthodontists
Sherif et al Screw- versus Cement-Retained Implant-Supported Fixed Restorations

Table 1 Inclusion and exclusion criteria case series or case reports were excluded. Three independent
evaluators assessed the studies produced from the database
Inclusion criteria
searches. After each step in the process of deletions (by title, ab-
Randomized clinical trials (including preference RCTs) stract, and full text), a Kappa statistic was calculated to evaluate
Prospective and retrospective cohort studies with patient recall interexaminer agreement.
Written in English The evaluators viewed the authors or titles. Studies that in-
Minimum 12 months of functional loading cluded insufficient information in the title were marked to be
Including healthy patients
retrieved for abstract review. From these abstracts, articles with
Method of intervention: Screw- or cement-retained restoration
insufficient information to merit their exclusion were retrieved
Exclusion criteria for full-text review. Two clinical academicians reviewed all
Immediate restoration studies set to be included at each. We defined clinical academi-
Clinical reports cians as full-time faculty members. Those studies selected for
Technique articles inclusion then underwent validity assessment and data retrieval.
Mean functional loading time less than 12 months Validity assessment was completed to ascertain the quality
Data not divisible into screw- and cement-retained cohorts of the study and was done primarily on the basis of follow-
Acrylic veneering of prosthesis∗ up and allocation concealment. Follow-up was assessed by
Cantilevers∗ thoroughness and completeness and any explanation for loss
Screw-retained crowns not having an objective torque control method. to follow-up (selection bias). Allocation concealment was as-
Hand tightening was excluded∗ sessed as described in the Cochrane Handbook for Systematic
Lingual set screws∗
Reviews of Interventions (v4.2.5), and was categorized as ad-
∗ equate, unclear, or inadequate. Finally, those studies accepted
Not a strict criterion for exclusion, provided that data that pertained to can-
were grouped according to bias: 1) low risk of bias and 2) high
tilevers or acrylic veneering materials could be separated from relevant included
risk of bias (including preference randomized clinical trials).
data. If this data could not be separated, the study was excluded.

Data retrieval and analysis

6. “mean plaque index OR MPI”
7. “sulcular bleeding index OR SBI”
8. “ceramic fracture”
All selected articles contained well-defined inclusion and
exclusion criteria (Table 1). Following the electronic search,
all nondental articles or those that used evidence from either
Data retrieval: For each study, the date of publication, author,
journal implant hex type, retention method, cement type, num-
ber of implants and prosthesis, major and minor factors listed
above, and demographic details were recorded. Exposure and
outcome variables per implant were taken including how and
at what time points they were assessed.

Table 2 Articles selected for inclusion in systematic review

First author Journal Year Research design Level of evidence

Andersson6 IJP 1998 Prospective cohort 2

Becker8 JPD 1995 Prospective cohort 2
Duncan9 JOMI 2003 Prospective cohort 2
Heckmann10 COIR 2006 In vitro/vivo -
Jemt11 IJP 2000 Prospective cohort 2
Karl12 JOMI 2006 In vitro -
Karl13 JPD 2007 In vitro -
Karl14 JPD 2008 In vitro/vivo -
Keith15 JOMI 1999 In vitro -
Krenmeir16 JOMI 2002 Retrospective cohort 2
Levine17 JOMI 1999 Retrospective cohort 2
Levine18 JOMI 2002 Retrospective cohort 2
Levine19 JOMI 2007 Retrospective cohort 2
Mericske-Stern20 COIR 2001 Prospective cohort 2
Norton21 JOMI 2006 Retrospective cohort 2
Priest22 JOMI 1999 Retrospective cohort 2
Scheller23 JOMI 1998 Prospective cohort 2
Schropp24 J Clinical Perio 2005 RCT 1b
Schwarz-Arad25 JP 1999 Retrospective cohort 2
Singer26 JOMI 1996 Retrospective cohort 2
Vigolo27 JOMI 2004 Retrospective cohort 2
Weber28 COIR 2006 Prospective cohort 2
Zitzmann29 JPD 2000 Prospective cohort 2

Journal of Prosthodontics 23 (2014) 1–9 

C 2013 by the American College of Prosthodontists 3
Screw- versus Cement-Retained Implant-Supported Fixed Restorations Sherif et al

Table 3 Major outcome failures, by study

Exposure Mean Failure 95% 95%

time Major exp. rate CI CI
Study Implants (Years) failures time Test (per 100 years) (lower) (upper)

Andersson-c 1998 60 279 5.00 4.7 3.3 1.79 0.29 10.1

Becker-s 1995 22 63.5 1.00 2.9 0.8 1.57 0.28 8.97
Duncan-s 2003 37 111 0.00 3.0 1.3 0.00 0.00 0.08
Duncan-c 2003 37 111 0.00 3.0 1.3 0.00 0.00 0.10
Jemt-s 2000 159 477 10.00 3.0 5.7 2.10 0.37 11.95
Krenmeir-s 2002 22 65.6 1.00 3.0 0.8 1.52 0.27 8.58
Krenmeir-c 2002 93 277.5 4.00 3.0 3.3 1.44 0.31 6.68
Levine-s 1999 81 162 0.00 2.0 1.9 0.00 0.00 0.14
Levine-c 1999 76 152 0.00 2.0 1.8 0.00 0.00 0.13
Levine-s 2002 71 126 0.00 1.8 1.5 0.00 0.00 0.07
Levine-c 2002 600 1065 6.00 1.8 12.6 0.56 0.12 2.61
Levine-c 2007 500 958 1.00 1.9 11.5 0.10 0.02 0.48
Norton-c 2006 173 533.4 13.00 3.1 6.4 2.44 0.53 11.29
Priest-s 1999 51 509.5 1.00 10.0 6.1 0.20 0.03 1.20
Priest-c 1999 61 610 2.00 10.0 7.3 0.33 0.06 1.69
Scheller-c 1998 77 358 9.00 4.6 4.3 2.51 0.04 15.42
Schropp-s 2005 2 3.00 0.00 1.5 0.1 0.00 0.00 0.00
Schropp-c 2005 43 64.5 0.00 1.5 0.8 0.00 0.00 0.00
Schwarz-Arad-c 1999 72 143.5 2.00 2.0 1.7 1.39 0.27 7.19
Singer-c 1996 225 298.4 3.00 1.3 3.6 1.01 0.19 5.18
Vigolo-s 2004 62 407 4.00 6.6 4.9 0.98 0.17 5.53
Vigolo-c 2004 125 868.3 1.00 6.9 10.4 0.12 0.02 0.53
Weber-s 2006 93 279 0.00 3.0 3.3 0.00 0.00 0.06
Weber-c 2006 59 177 0.00 3.0 2.1 0.00 0.00 0.09
Zitzmann-s 2000 81 258.7 2.00 3.2 3.1 0.77 0.14 4.41
Total Summary 2882 8357.9 65.00 2.9 100 0.81 0.00 6.85

c: cement-retained; s: screw-retained.

Data analysis: Risk ratios were applied to dichotomous data Table 4 Poisson model for major failures
to estimate an intervention’s effects with a 95% confidence
Coefficient S.E. P-value
interval. Continuous outcomes used a combination of mean
differences and standard deviations. When studies used similar Screw − 0.51 0.82 0.54
outcome measures, a meta-analysis was performed. The risk RCT − 22.82 1.62 <0.01
ratios were combined if the data included were dichotomous, Retrospective 0.5 0.87 0.57
while mean differences were used for continuous data. Split- Published after 2000 − 1.5 0.84 0.09
mouth data were combined with data from parallel-group trials.
Depending upon the outcome of interest, the implant or associ- Statistically significant associations are listed in bold typeface.
ated restoration was the statistical unit. Random effects Poisson
models were used to analyze the failure and complication rates. for full-text review. After full-text review, there was significant
As count data are often overdispersed (variance larger than the agreement between the reviewers (k = 0.84). Final discussion
mean and/or containing a large number of zeros), the random ef- with the academicians resulted in the inclusion of 24 articles, of
fects represents this unobserved heterogeneity. For all analyses, which the authors were able to obtain information for 23 of the
a p-value ≤ 0.05 was considered statistically significant. articles (the author of one article did not respond to our request
for data).6-28 The included articles are listed in Table 2.
Results
Rate of major failure
After deletion of duplicates, the database searches resulted in
a combined 577 publications to be evaluated by title. Three in- Data were extracted from 17 papers (Table 3). For each paper
dependent reviewers had moderate agreement for inclusion of with both retention types, the data were separated by retention
articles by title (k = 0.64), with 295 articles being chosen after type, yielding 25 studies for analysis. After adjusting for study
discussion with two academicians. After the reviewers evalu- type, retention type, and publication year (before or after 2000),
ated the abstracts of these articles, moderate agreement was a random effects Poisson model was fitted to the counts of major
achieved on article selection (k = 0.47). Subsequent discussion failure, with total exposure time as the offset. The estimated
with the academicians resulted in the inclusion of 81 articles coefficients are shown in Table 4. Randomized controlled trial
4 Journal of Prosthodontics 23 (2014) 1–9 
C 2013 by the American College of Prosthodontists
Sherif et al Screw- versus Cement-Retained Implant-Supported Fixed Restorations

Figure 2 Estimated major failure rates and confi-

dence intervals for each paper, where the dashed
line is the overall rate. (c: cement-retained; s:
screw-retained).

(RCT) studies had a significantly lower failure rate (p < 0.01) Table 6 Poisson model for screw loosening
than both prospective and retrospective studies, with studies
Coefficient S.E. p-value
published after 2000 having a marginally lower failure rate
(p-value 0.09). Internal hex 1.33 1.43 0.39
The summary of the studies and the estimated failure rates RCT −22.12 2.34 <0.01
are listed in Figure 2. The major failure rate was 0.87 per 100 Retrospective 0.80 1.16 0.95
years (95% CI: 0.00, 11.03) for studies with cement retention Published after 2000 −0.84 1.43 0.58
type, and 0.71 per 100 years (95% CI: 0.00, 15.65) for studies
with screw retention. The difference in failure rates between Statistically significant associations are listed in bold typeface.
these two retention types was not statistically significant (p =
0.54). The overall failure rate between these two retention types
was 0.81 per 100 years (95% CI: 0, 6.85). Poisson model was fit to the counts of screw loosening, with
total exposure time as the offset. The model coefficients are
shown in Table 6. RCTs had a significantly lower rate (p <
Rate of screw loosening
0.01) of screw loosening than both prospective and retrospec-
Data were extracted from 11 papers and are summarized in tive studies. The summary of the studies and the failure rates
Table 5. After adjusting for study type, internal/external hex, are presented in Figure 3. The overall failure rate of screw
and publication year (before or after 2000), a random effects loosening was 3.66 per 100 years (95% CI: 0.0, 49.37).

Table 5 Screw loosening outcomes, by study

Mean exp. % of Cumulative Failure 95% 95%

Exposure time exp. time in screw rate CI CI
Author Year Implants time (years) all studies loosening (per 100 years) (lower) (upper)

Becker 1995 22 63.5 2.9 2.6 8.00 12.6 0.63 259.72

Duncan 2003 37 111 3.0 4.5 9.00 8.11 0.35 185.94
Jemt 2000 159 477 3.0 19.4 4.00 0.84 0.04 17.08
Krenmeir 2002 22 65.6 3.0 2.7 0.00 0.00 0.02 8.72
Levine 1999 81 162 2.0 6.6 7.00 4.32 0.13 138.7
Levine 2002 71 126 1.8 5.1 13.00 10.32 0.51 209.77
Priest 1999 51 509.5 10.0 20.7 8.00 1.57 0.07 36.01
Schropp 2005 2 3.00 1.5 0.1 0.00 0.00 0.00 0.00
Vigolo 2004 62 407 6.6 16.5 2.00 0.49 0.02 15.79
Weber 2006 93 279 3.0 11.3 0.00 0.00 0.00 11.4
Zitzmann 2000 81 258.7 3.2 10.5 0.00 0.00 0.00 2.87
Total - 681 2462.3 3.6 100 53.00 3.66 0.00 49.37

Journal of Prosthodontics 23 (2014) 1–9 

C 2013 by the American College of Prosthodontists 5
Screw- versus Cement-Retained Implant-Supported Fixed Restorations Sherif et al

Figure 3 Estimated screw loosening rates and confidence

intervals for each paper, where the dashed line is the
overall rate.

Table 7 Loss of retention outcomes, by study ture, with total exposure time as the offset. The model coeffi-
cients are shown in Table 10. RCTs had a significantly lower
Coefficient S.E. p-value
rate (p < 0.01) of porcelain fracture than both prospective and
Intercept 4.44 0.96 < 0.0001 retrospective studies did. The estimated porcelain fracture rate
RCT − 22.12 2.34 < 0.0001 and confidence intervals for each paper are listed in Figure 5.
Retrospective 0.80 1.16 0.95
Internal hex 1.33 1.43 0.39
Discussion
After 2000 − 0.84 1.43 0.58
This systematic review included 3084 implant and five in vitro
Statistically significant associations are listed in bold typeface. studies from a total of 23 publications, which included one
randomized controlled trial,24 and eight prospective and nine
Table 8 Poisson model for loss of retention retrospective cohort studies.6,8-29 Of the 81 articles selected
for full text review, 58 were deleted. Seventeen were deleted
Coefficient S.E. p-value
due to bias, ten were deleted due to lack of follow-up, and 31
Internal hex 0.79 0.84 0.37 were deleted as they were not relevant to the meta-analysis.
RCT 3.5 1.63 0.06 Interevaluator agreement was high, with k ranging from mod-
Retrospective 1.13 1.10 0.33 erate to almost perfect agreement for the three stages of the
Published after 2000 − 0.66 0.72 0.38 systematic deletion of publications from the study.
In the presence of an RCT meeting of the inclusion criteria,
a stratification of levels 1 and 2 evidence was used. Though
Rate of decementation language was not an exclusion criterion in the systematic dele-
tion of publications, no non-English text studies were selected
Data were extracted from 14 papers with cement-retention-type
for full-text review, as we could not reliably interpret such
cohorts and are summarized in Table 7. After adjusting for study
literature to see if it fit our strict inclusion criteria. An empir-
type, internal/external hex, and publication year (before or after
ical study found that this exclusion was not a disadvantage,
2000), a random effects Poisson model was fit to the counts of
as studies published in a language other than English tended
loss of retention, with total exposure time as the offset. The
to overestimate the treatment effect by 10%,30 while unpub-
model coefficients are shown in Table 8. RCT studies had a
lished studies would underestimate the intervention effect by
marginally higher rate (p = 0.06) of retention loss than both
the same percentage. This same study demonstrated that papers
prospective and retrospective studies. The overall failure rate
not indexed in MEDLINE overestimate the therapeutic effect
from decementation was 2.54 per 100 years (95% CI: 0, 16.30,
by 5%. This study concluded that the quality of the trial is more
Fig 4).
important than the reporting and dissemination of the informa-
tion gathered from the trial in terms of source of bias. Between
Porcelain fracture rate
cement- and screw-retained crowns, there was no significant
Data were extracted from 25 cohorts in 17 papers, as sum- difference in the actual major failure outcome rate (0.71 vs.
marized in Table 9. After adjusting for retention type, study 0.87/ 100 years, respectively).
type, and publication year (before or after 2000), the random When performing the data extraction, some manuscripts were
effects Poisson model was fit to account for porcelain frac- unable to be analyzed thoroughly for follow-up or allocation

6 Journal of Prosthodontics 23 (2014) 1–9 

C 2013 by the American College of Prosthodontists
Sherif et al Screw- versus Cement-Retained Implant-Supported Fixed Restorations

Figure 4 Estimated loss of retention rates for

each paper, where the dashed line is the
overall rate.

of bias. As such, the authors of ten papers were contacted by
e-mail to elaborate on their data. An example of a common
question was, “to the nearest 6-month interval, when did the
porcelain fracture occur in the screw-retained group?” or, “how
many times did the prosthetic screw of implant three become
loose?” The authors of nine papers replied with answers that
allowed the publication to undergo complete data extraction.
One author did not reply after three e-mails, and additional
correspondence was exchanged with that author’s secretary.
This study could not be included as it was not known how

Table 9 Porcelain fracture outcomes, by study

Mean % of Failure 95% 95%

Porcelain exp. exp. time rate CI CI
Author Year fracture time all studies (per 100 years) (lower) (upper)

Andersson 1998 2.00 4.7 3.3 0.72 0.05 9.57

Becker 1995 0.00 2.9 0.8 0 0 0.34
Duncan 2003 0.00 3 1.3 0 0 0.48
Duncan 2003 0.00 3 1.3 0 0 0.35
Jemt 2000 9.00 3 5.7 1.89 0.15 23.83
Krenmeir 2002 1.00 3 0.8 1.52 0.12 19.07
Krenmeir 2002 2.00 3 3.3 0.72 0.07 7.84
Levine 1999 0.00 2 1.9 0 0 0.15
Levine 1999 0.00 2 1.8 0 0 0.1
Levine 2002 0.00 1.8 1.5 0 0 0.2
Levine 2002 2.00 1.8 12.6 0.19 0.02 2.04
Levine 2007 0.00 1.9 11.5 0 0 0.1
Norton 2006 13.00 3.1 6.4 2.44 0.02 26.53
Priest 1999 0.00 10 6.1 0 0 0.15
Priest 1999 1.00 10 7.3 0.16 0.01 1.92
Scheller 1998 7.00 4.6 4.3 1.96 0.15 26.1
Schropp 2005 0.00 1.5 0.1 0 0 0
Schropp 2005 0.00 1.5 0.8 0 0 0
Schwarz-Arad 1999 1.00 2 1.7 0.7 0.06 8.18
Singer 1996 2.00 1.3 3.6 0.67 0.06 7.87
Vigolo 2004 0.00 6.6 4.9 0 0 0.2
Vigolo 2004 0.00 6.9 10.4 0 0 0.1
Weber 2006 0.00 3 3.3 0 0 0.46
Weber 2006 0.00 3 2.1 0 0 0.34
Zitzmann 2000 0.00 3.2 3.1 0 0 0.31
SUMMARY 40.00 2.9 100 0.46 0 13.51

Journal of Prosthodontics 23 (2014) 1–9 

C 2013 by the American College of Prosthodontists 7
Screw- versus Cement-Retained Implant-Supported Fixed Restorations
Table 10 Poisson model for porcelain fracture
Coefficient S.E.
Screw − 1.07 0.85
RCT − 22.52 1.68
Retrospective − 0.33 0.90
Published after 2000 − 0.64 0.87
Statistically significant associations are listed in bold typeface.
many restorations were screw- or cement-retained, or in which
cohort the major failure occurred. Exclusion of this article was
effected to ensure that strict inclusion criteria were maintained
and that data amenable to analysis were available.
The process of contacting authors after reading full texts led
to one extra article becoming eligible for evaluation in the meta-
analysis. Levine et al were contacted regarding their 1999 and
2002 papers, and proposed their 2007 article that had not been
retrieved by any of the nine database searches.17-19 All three
evaluators selected this article for data extraction.
The study had strict exclusion criteria to ensure validity when
abstracts and full texts were analyzed. Such examples include
excluding restorations that included a cantilever or contained
an acrylic instead of porcelain veneering material. This allowed
data from certain studies to be analyzed while discounting non-
ideal restorations. By comparison, the initial deletions of titles
were evaluated liberally to ensure that no article was erro-
neously deleted before more information was available on their
content.
While no studies were deleted that were subsequently rein-
troduced to the study by either analysis of the references, hand
searching, or on recommendations of contacted authors, one
study was newly introduced at the full-text review stage. On
contacting Levine et al to gain additional data on two studies,
as is mandated in the Cochrane Handbook for Systematic Re-
views of Interventions, another paper was recommended for
analysis, as noted above.17-19 As this article was not in the orig-
inal 26,582 articles produced from the database search, it was
not affected by the exclusion criteria. Possible reasons for the
8 Journal of Prosthodontics 23 (2014) 1–9 
Sherif et al
study’s omission from PubMed and MEDLINE may include
keyword indexing and the number of phrases included in the
P-value
MeSH search. To ensure validity that no data were excluded
0.22 based on the search term limitation (e.g., keep screw/cement),
<0.01 the authors selected three random groups of 20 articles each
0.72 prior to elimination of studies that did not include these terms.
0.47 The exclusion criteria were then applied to these titles and
abstracts. All 60 articles were deleted as they contained in-
formation listed in the exclusion criteria, thus aiding in the
assessment that the study is valid in the articles deleted.
Major outcomes included loss of the crown or implant. The
difference between the two cohorts was not significant with an
overall failure rate of 0.81 per 100 years (p = 0.54; 95% CI: 0,
6.85). When evaluated by individual cohorts, the major failure
rate was 0.87 per 100 years (95% CI: 0.00, 11.03) for studies
with cement retention type, and 0.71 per 100 years (95% CI:
0.00, 15.65) for studies with screw retention type. The 95%
confidence intervals were larger than that of the combined rate
due to the smaller sample size in each separate group.
Possible reasons for the lower but nonsignificant major fail-
ure rate of screw-retained crowns include the experience of the
operator and clinical indications for use of the cement-retained
crown. Cement-retained crowns have more in common with
regular fixed prosthodontics than do screw-retained restora-
tions, and as such have a wider appeal to practitioners of all
experience levels. It may be hypothesized that screw-retained
restorations are still preferred by more specialists than gener-
alists, and thus are used less frequently and with more special-
ist training than cement-retained units. Second, screw-retained
restorations are held to stricter criteria in the treatment-planning
phase.
The minor outcomes included screw loosening, decementa-
tion, and porcelain fracture. There were no significant differ-
ences between the two cohorts for all three parameters. Screw
loosening occurred 3.66 times per 100 years, while decemen-
tation occurred 2.54 times per 100 years. Porcelain fracture
was not significantly lower in the screw-retained group. Both
methods of retention are successful in the restoration of par-
tially edentulous patients. It was expected that the porcelain
Figure 5 Estimated porcelain failure rates and asso-
ciated confidence intervals for each paper, where the
dashed line is the overall rate (c: cement-retained; s:
screw-retained).
C 2013 by the American College of Prosthodontists
Sherif et al
fracture rate would be higher in the screw-retained group due
to the screw access hole and occlusion not being centered in the
fossa.12 It could be hypothesized that the increased loss of reten-
tion in the screw-retained patients acted to prevent subsequent
porcelain fracture; however, all differences were nonsignificant.
Conclusion
Overall, the major failure rate is 0.81 over 100 years. While this
was slightly more for the cement-retained group (0.87 per 100
years compared to 0.71 per 100 years for studies with screw
retentions), the difference is not statistically significant. The
minor outcomes included screw loosening, decementation, and
porcelain fracture. There were no significant differences be-
tween the two cohorts for all three parameters. This is impor-
tant data as it shows that screw retention methods are equally
suitable for the partially edentulous patient, although cement-
retained restorations are more frequently used. Future research
should focus on clinical and microbiological enhancement of
cement- and screw-retained implant therapy.
References
1. Hobkirk JA, Schwab J: Mandibular deformation in subjects with
osseointegrated implants. Int J Oral Maxillofac Implants
1991;6:319-328
2. Adell R, Leckholm U, Rockler B, et al: A 15-year study of
osseointegrated implants in the treatment of the edentulous jaw.
Int J Oral Surg 1981;10:387-416
3. Albrektsson T: A multicenter report of osseointegrated oral
implants. J Prosthet Dent 1988; 60:75-84
4. Adell R, Eriksson B, Leckholm U, et al: A long-term follow-up of
osseointegrated implants in the treatment of the treatment totally
edentulous jaws. Int J Oral Maxillofac Implants 1990;5:347-359
5. Lewis S, Beumer J III, Hornburg W, et al: The “UCLA”
abutment. Int J Oral Maxillofac Implants 1988;3:183-189
6. Andersson B, Odman P, Lindvall AM, et al: Cemented single
crowns on osseointegrated implants after 5 years: results from a
prospective study on CeraOne. Int J Prosthodont
1998;11:212-218
7. Taylor TD, Agar JR: Twenty years of progress in implant
prosthodontics. J Prosthet Dent 2002;88:89-95
8. Becker W, Becker BE: Replacement of maxillary and
mandibular molars with single endosseous implant restorations:
a retrospective study. J Prosthet Dent 1995;74:71-75
9. Duncan JP, Nazarova E, Vogiatzi T, et al: Prosthodontic
complications in a prospective clinical trial of single-stage
implants at 36 months. Int J Oral Maxillofac Implants
2003;18:561-565
10. Heckmann SM, Karl M, Wichmann MG, et al: Loading of bone
surrounding implants through three-unit fixed partial denture
fixation: a finite-element analysis based on in vitro and in vivo
strain measurements. Clin Oral Implants Res 2006;17:345-350
11. Jemt T, Henry P, Linden B, et al: A comparison of laser-welded
titanium and conventional cast frameworks supported by
implants in the partially edentulous jaw: a 3-year prospective
multicenter study. Int J Prosthodont 2000;13:282-288
12. Karl M, Graef F, Taylor TD, et al: In vitro effect of load cycling
on metal-ceramic cement- and screw-retained implant
restorations. J Prosthet Dent 2007;97:137-140
13. Karl M, Graef F, Wichmann MG, et al: The effect of load cycling
on metal ceramic screw-retained implant restorations with
Journal of Prosthodontics 23 (2014) 1–9 
C 2013 by the American College of Prosthodontists
Screw- versus Cement-Retained Implant-Supported Fixed Restorations
unrestored and restored screw access holes. J Prosthet Dent
2008;99:19-24
14. Karl M, Winter W, Taylor TD, et al: Fixation of 5-unit
implant-supported fixed partial dentures and resulting bone
loading: a finite element assessment based on in vivo strain
measurements. Int J Oral Maxillofac Implants 2006;21:756-
762
15. Keith SE, Miller BH, Woody RD, et al: Marginal discrepancy of
screw-retained and cemented metal-ceramic crowns on implants
abutments. Int J Oral Maxillofac Implants 1999;14:369-378
16. Krennmair G, Schmidinger S, Waldenberger O: Single-tooth
replacement with the Frialit-2 system: a retrospective clinical
analysis of 146 implants. Int J Oral Maxillofac Implants
2002;17:78-85
17. Levine RA, Clem DS, Wilson TG, et al: Multicenter
retrospective analysis of the ITI implant system used for
single-tooth replacements: results of loading for 2 or more years.
Int J Oral Maxillofac Implants 1999;14:515-520
18. Levine RA, Clem D, Beagle J, et al: Multicenter retrospective
analysis of the solid-screw ITI implant for posterior single-tooth
replacements. Int J Oral Maxillofac Implants 2002;17:
550-556
19. Levine RA, Ganeles J, Jaffin RA, et al: Multicenter retrospective
analysis of wide-neck dental implants for single molar
replacement. Int J Oral Maxillofac Implants 2007;22:736-
742
20. Mericske-Stern R, Grutter L, Rosch R, et al: Clinical evaluation
and prosthetic complications of single tooth replacements by
non-submerged implants. Clin Oral Implants Res
2001;12:309-318
21. Norton MR: Multiple single-tooth implant restorations in the
posterior jaws: maintenance of marginal bone levels with
reference to the implant-abutment microgap. Int J Oral
Maxillofac Implants 2006;21:777-784
22. Priest G: Single-tooth implants and their role in preserving
remaining teeth: a 10-year survival study. Int J Oral Maxillofac
Implants 1999;14:181-188
23. Scheller H, Urgell JP, Kultje C, et al: A 5-year multicenter study
on implant-supported single crown restorations. Int J Oral
Maxillofac Implants 1998;13:212-218
24. Schropp L, Kostopoulos L, Wenzel A, et al: Clinical and
radiographic performance of delayed-immediate single-tooth
implant placement associated with peri-implant bone defects. A
2-year prospective, controlled, randomized follow-up report.
J Clin Periodontology 2005;32:480-487
25. Schwartz-Arad D, Samet N: Single tooth replacement of missing
molars: a retrospective study of 78 implants. J Periodontol
1999;70:449-454
26. Singer A, Serfaty V: Cement-retained implant-supported fixed
partial dentures: a 6-month to 3-year follow-up. Int J Oral
Maxillofac Implants 1996;11:645-659
27. Vigolo P, Givani A, Majzoub Z, et al: Cemented versus
screw-retained implant-supported single-tooth crowns: a 4-year
prospective clinical study. Int J Oral Maxillofac Implants
2004;19:260-265
28. Weber HP, Kim DM, Ng MW, et al: Peri-implant soft-tissue
health surrounding cement- and screw-retained implant
restorations: a multi-center, 3-year prospective study. Clin Oral
Implants Res 2006;17:375-379
29. Zitzmann NU, Marinello CP: Treatment outcomes of fixed or
removable implant-supported prostheses in the edentulous
maxilla. Part II: clinical findings. J Prosthet Dent
2000;83:434-442
30. Egger M, Enrahim S, Smith GD: Where now for meta-analysis?
Int J Epidemiol 2002;31:1-5
9