Review Article
Do Metal Post–retained Restorations Result in More Root
Fractures than Fiber Post–retained Restorations?
A Systematic Review and Meta-analysis
Fabr
ıcio Eneas Diniz Figueiredo, DDS,* Paulo Ricardo Saquete Martins-Filho, PhD,†
and Andr
e Luis Faria-e-Silva, PhD*
Abstract
Introduction: Teeth requiring endodontic treatment
commonly have compromised a coronal tooth structure
that often requires the use of an intraradicular post to
retain the coronal restoration. Although usually success-
ful, catastrophic failures requiring extraction have been
reported in the literature. The aim of this systematic re-
view was to analyze clinical trials and cohort studies that
evaluated the incidence rate of root fractures in post-
retained restorations. The hypothesis was that the inci-
dence rate related to the use of metal posts was higher
than that of fiber posts. Methods: A MEDLINE search
for clinical studies reporting the incidence of root frac-
tures of restorations retained with fiber posts or metal
posts of endodontically treated teeth with a more than
5-year follow-up was conducted from inception to
January 2014. Seven randomized clinical trials and 7
cohort studies were included. Results: The pooled sur-
vival rate was 90% (95% confidence interval, 85.5–
93.3) for metal-based posts and 83.9% (95% confidence
interval, 67.6–92.8) for fiber-reinforced posts. The over-
all incidence rate of root fractures (catastrophic failures)
was similar between metal and fiber posts. Prefabri-
cated metal posts and carbon fiber posts had a 2-fold in-
crease in the incidence rate of root fractures compared
with cast metal posts and glass fiber posts, respectively.
Conclusions: The results of this study did not show sig-
nificant differences for root fracture incidence between
metal- and fiber posts. However, the studies included
in this review presented a high risk of bias, and further
well-designed clinical studies are required to confirm
these findings. (J Endod 2015;41:309–316)
Key Words
Dental restoration failure, endodontically treated teeth,
meta-analysis, post and core technique
From the Departments of *Dentistry and †Health Education,
Universidade Federal de Sergipe, Sergipe, Brazil.
Address requests for reprints to Dr Andr
e Luis Faria-e-Silva,
Universidade Federal de Sergipe, Hospital Universit
ario, Depar-
tamento de Odontologia, Rua Cl
audio Batista, s/n Bairro San-
at
orio, Aracaju, Sergipe, Brazil, CEP 49060-100. E-mail
address: fariaesilva.andre@gmail.com
0099-2399/$ - see front matter
Copyright ª 2015 American Association of Endodontists.
http://dx.doi.org/10.1016/j.joen.2014.10.006
JOE — Volume 41, Number 3, March 2015
P refabricated fiber-reinforced posts have become more popular than cast posts for
clinical usage in recent years. This may be caused by enhanced esthetics and a
reduction in treatment time. Another factor may be that the elastic modulus between
fiber posts and dentin is similar, which has been related to a reduced incidence of
root fractures (1, 2). Several in vitro studies have shown that because of their high
elastic modulus, metal posts concentrate stress on the root and promote a higher
incidence of root fractures compared with fiber posts (3). In contrast, a recent study
with finite element analysis has shown that the use of fiber posts resulted in higher stress
on root structures compared with metal posts, primarily when the bonding between the
fiber posts and the root canal surface failed (ie, debonding) (4). However, the same
study showed that fiber post–restored roots were less prone to fracture because the
risk of fracture of the core and/or post is higher than that of the root.
Clinically, a reduced incidence of catastrophic failures, which does not permit the
replacement of the restoration, can be more important than the survival rate of the
restoration. However, the aim of most reviews on intraradicular posts has been to eval-
uate the survival rate. Thus, the aim of this systematic review was to analyze the outcome
of clinical trials and cohort studies evaluating the post-retained restorations regarding
the incidence rate of root fractures. The hypothesis is that the incidence rate of root
fractures related to the use of metal posts is higher than that of fiber posts. This study
also aimed to evaluate the clinical performance of the systems regarding their survival
rate.
Materials and Methods
A protocol of this systematic review was designed a priori and was registered in the
PROSPERO database (registration number CRD 42014007423).
Eligibility Criteria
Clinical studies reporting the incidence of root fractures of restorations retained
with fiber-reinforced composite posts or metal posts of endodontically treated teeth
were considered eligible for inclusion in this systematic review. Articles were selected
if they met the following criteria: randomized clinical trials (RCTs) and cohort studies
comparing the incidence of root fractures of metallic and fiber-reinforced composite
posts, RCTs and cohort studies in which 1 arm reported the incidence of root fracture
of 1 system, and RCTs and cohort studies with a mean/median of follow-up time of
5 years.
Studies from which we were unable to extract data for the outcomes of interest,
in vitro studies, reviews, and studies with a mean/median of follow-up time of less
than 5 years were excluded. Whenever more than 1 publication reported results for
the same group of patients, we included only the report containing the most compre-
hensive data to avoid the duplication of information.
Search Strategy
A MEDLINE search for RCTs and cohort studies was conducted from inception (no
limit regarding the year of publication) to January 2014 using the following key terms:
(‘‘nonvital tooth’’ OR ‘‘devitalized tooth’’ OR ‘‘pulpless tooth’’ OR ‘‘endodontically
Root Fracture of Post-retained Restorations 309
Review Article
Figure 1. A flowchart for the selection of systematic reviews.
treated tooth’’) AND (‘‘fiber post’’ OR ‘‘metallic post’’ OR ‘‘cast dowel’’
OR ‘‘dowel’’ OR ‘‘metal post’’ OR ‘‘carbon-fiber post’’ OR ‘‘glass-fiber
post’’ OR ‘‘quartz-fiber post’’ OR ‘‘fiber-reinforced post’’ OR ‘‘post
core systems’’ OR ‘‘post and core technique’’) AND (exp cohort studies
OR cohort OR controlled clinical trial OR epidemiologic methods OR
clinical trial). The search also included a hand search of cross-
references from original articles and reviews to identify additional
studies that could not be located in the MEDLINE database. In addition,
dissertations and theses were searched in the OpenThesis database us-
ing a search strategy similar to the one applied in MEDLINE. No lan-
guage or publication year criteria were imposed.
Data Extraction and Outcomes
Two independent reviewers screened the search results and iden-
tified studies that were potentially relevant based on the article titles and
abstracts. Relevant studies were read in full and selected according to
310 Figueiredo et al.
the eligibility criteria. Data concerning cohort studies were extracted
independently by 2 reviewers. Disagreement between the 2 reviewers
was solved either by consensus or a third reviewer.
Root fracture leading to tooth extraction was considered a cata-
strophic failure and defined as the primary outcome. Noncatastrophic
failures were defined as the secondary outcome and included endodon-
tic failures (ie, failures related exclusively to the root canal therapy, but
the restoration remained intact), crown dislodgement, post debonding,
and post/core fractures (5, 6).
Assessment of Risk of Bias
The Cochrane Risk of Bias Tool was used to assess the quality of
the study methodology of eligible RCTs. The study quality was assessed
independently by 2 reviewers using the following 7 criteria: random
sequence generation, allocation concealment (selection bias), blinding
of participants and personnel (performance bias), blinding of outcome
JOE — Volume 41, Number 3, March 2015

TABLE 1. Summary Characteristics of Included Studies
Randomized clinical trials
Eligible studies
No. of unique studies
Median (IQR) follow-up (y)
Participants
Total no. of participants
Median (IQR) no. of participants
% female
Post system
Metal-based posts (no. of studies)
Fiber-reinforced posts (no. of studies)
Post years of follow-up*
Outcome (no. of events)†
Root fractures
Noncatastrophic failures
IQR, interquartile range.
*Estimated from the number of transplanted patients and the mean follow-up time.
†
Several studies provided data on multiple outcomes of interest.
assessment (detection bias), incomplete outcome data (attrition bias),
selective reporting (reporting bias), and other bias (absence of
description regarding tooth location or the amount of remaining coro-
nal structure). The response for each criterion was reported as low risk
of bias, high risk of bias, and unclear risk of bias (7).
The cohort studies were assessed using a modified version of the
Newcastle-Ottawa Scale (8) and included evaluations of the representa-
tiveness of the sample size, selection of a comparison group (whether
participants were drawn from the same source, eg, same institution or
database), ascertainment of exposure by secure records, outcome of
interest not present at the start of the study, confounders or others fac-
tors used to match or control for in analysis, outcome assessment by
clinical and radiographic examination, adequate follow-up period for
outcome of interest ($5 years), and loss to follow-up unlikely to intro-
duce bias. These items were not merged into a quality score. Instead, the
relevant information for each domain was tabulated to allow for greater
transparency. The Newcastle-Ottawa Scale was also assessed by 2 re-
viewers, and disagreement was solved by consensus or a third reviewer.
Data Synthesis and Analysis
The incidence rate of root fractures and noncatastrophic failures
was calculated by dividing the number of events by the period of risk for
all included posts during the study period per 1000 post years of follow-
up. If the total number of posts years was not reported, it was calculated
by multiplying the number of posts under follow-up by the mean dura-
tion of follow-up. Confidence intervals (CIs) were based on the normal
approximation and calculated using the number of posts and post years
of follow-up. A constant 0.5 continuity correction was used for studies
with 0 events (9).
A meta-analysis of the crude survival rates with no adjustment for
the follow-up duration was performed by dividing the number of surviv-
ing posts by the total number of posts included during the study period.
A Kaplan-Meier analysis could not be performed because most of the
reports did not provide a precise follow-up period for each patient.
CIs were calculated using normal approximation. When the crude sur-
vival rate in a study was reported to be 100%, we used a continuity
correction with the crude survival rate set to 99%.
Statistical heterogeneity was investigated using the Cochran Q test
and I2 index. For the Q statistic, a P value <.10 was used as an indication
of interstudy variability (10). I2 was interpreted as the proportion of to-
tal variation across studies caused by the variability, and a value >75%
JOE — Volume 41, Number 3, March 2015
Review Article
Observational cohort studies All studies
7 7 14
6.7 (5.9–7.8) 10.0 (8.6–10.0) 8.2 (6.6–10.0)
874 2328 3202
91.0 (59.0–151.0) 99.0 (78.5–515.0) 95.0 (75.3–173.8)
54.9 68.4 65.2
4 4 8
5 3 8
7427 34 294 41 721
8 54 62
54 413 467
was considered a measure of high heterogeneity (11). Subgroup ana-
lyses according to the study design (RCT  cohort), sample size
(<100 posts  $100 posts), and post system (metal post–restored
roots  cast metal post–restored roots, reinforced carbon fiber
post–restored roots  reinforced glass fiber post–restored roots)
were conducted to explore any potential source of heterogeneity be-
tween studies. For a meta-analysis of the survival rate, a stratified anal-
ysis according to the follow-up time was performed.
The pooled estimate of the rates was calculated using the
fixed-effects or random-effects models depending on the between-
study heterogeneity. In the absence of high heterogeneity, a fixed-
effects model using the Mantel-Haenszel method was selected to pool
the data, assuming that there was a common effect size and that the dif-
ferences between single results were caused by chance. When high het-
erogeneity was observed, the pooled estimates and the corresponding
95% CIs were calculated based on the random-effects model using the
DerSimonian-Laird method. In the random-effects meta-analysis model,
the rates for individual studies were assumed to vary around pooled es-
timates (12). The forest plot was used to present the pooled estimates and
the 95% CIs graphically. Each study was represented by a square in the
plot that was proportional to the study’s weight in the meta-analysis. Two-
sided P values <.05 were considered statistically significant.
A funnel plot was created by plotting the individual estimates in log
units against the standard error. To assess publication bias, contour-
enhanced funnel plots were examined, and their symmetry was tested
using the Begg and Mazumbar rank correlation with continuity correc-
tion (13). ‘‘Leave-one-out’’ sensitivity analysis was conducted by omit-
ting 1 study at a time and examining the influence of each individual
study on the pooled effect size (14).
Results
Search Results and Study Characteristics
The literature search resulted in 255 articles, 20 of which were
defined as potentially relevant to the current analysis. Of these, 9
were excluded in the subsequent detailed assessments for the reasons
shown in Figure 1. No dissertation or thesis was retrieved from the
OpenThesis database. Three other articles were included after search-
ing the references and adding these articles to the remaining 11 studies,
resulting in a total of 14 studies (6, 15–27) that met our eligibility
criteria and were included in the meta-analysis. The 14 studies (7
RCTs and 7 cohort studies) included in the present analysis were
Root Fracture of Post-retained Restorations 311
Review Article
TABLE 2. Risk of Bias of Included Studies
Quality of cohorts using the Newcastle-Ottawa quality scale
Selection Outcome
Outcome not
Selection of Ascertainment present at Confounders
Representative a comparison of exposure by beginning and other Assessment Was follow-up Follow-up
Cohort, year sample size group secure records of study factors of outcome long enough? rate $90%
Naumann et al, 2012 + + + + +
Go
mez-Polo et al, 2010 + + + + +
Hikasa et al, 2010 + + + + + +
Jung et al, 2007 + + + + + +
Ferrari et al, 2007 + + + + + +
Segerstro€ m et al, 2006 + + + + +
Weine et al, 1991 + + + + +

Quality of RCTs using Cochrane risk of bias tool

Selection Performance Detection Attrition Reporting
Random Blinding of Blinding of
Randomized clinical sequence Allocation participant outcome Incomplete Selective
trial, year generation concealment and personnel assessment outcome data reporting Other
Sterzenbach et al, 2012 Unclear risk Unclear risk High risk Low risk High risk Low risk Low risk
Ferrari et al, 2012 Unclear risk Unclear risk High risk High risk High risk Low risk Low risk
Signore et al, 2009 High risk Low risk High risk High risk High risk Low risk High risk
Creugers et al, 2005 Unclear risk Unclear risk High risk Unclear risk Unclear risk Low risk High risk
Mannocci et al, 2005 Unclear risk Unclear risk High risk High risk High risk Low risk Low risk
Ellner et al, 2003 High risk Unclear risk High risk Unclear risk Low risk Low risk High risk
King et al, 2003 Unclear risk Unclear risk High risk High risk High risk Low risk High risk

published between 1991 and 2012. The pooled population was
composed of 3202 participants and 4752 posts, with a follow-up
time of 41,721 post years. Sixty-two root fractures and 467 noncata-
strophic failures were diagnosed during the study period (Table 1).
Risk of Bias
Randomization and allocation concealment procedures were
inadequate or unclear in the trials included in the systematic review.
Failures in randomization and allocation concealment allow the clini-
cian to predict the upcoming treatment allocation, thereby leading to
selection bias. In addition, except in the study performed by Sterzen-
bach et al (26), neither the patients nor the investigators were blinded
to the treatment received, and the trials were classified as having a high
risk of performance and detection bias. Only 1 study (16) had no attri-
tion bias, and a low risk of reporting bias was observed for all trials. The
absence of descriptions regarding the amount of coronal remaining
and/or tooth location was also considered a bias. The quality assess-
ment of the included trials is shown in detail in Table 2.
The risk of bias among cohort studies is reported in Table 2. The
sample size was considered representative in all cohort studies, but in
only 3 studies (17, 19, 20) was a comparison group selected. The
ascertainment of exposure through secure records and demonstration
that the outcome of interest was not present at the start of the study
were observed for all cohorts. The methods used to control for
confounding were not described, and a high risk of performance bias
was observed. However, outcome assessment by clinical and
radiographic examination was described for most studies. Although
the follow-up time was long enough to produce root fractures and non-
catastrophic failures, only 2 cohort studies (19, 24) described that a loss
to follow-up was unlikely to introduce bias.
Incidence Rate of Root Fracture
Data on the incidence rate of root fractures for metal-based posts
were extractable from 4 RCTs (15, 16, 21, 26) and 4 cohort studies
(18–20, 27). The pooled event rate per 1000 post years was 5.13
(95% CI, 4.05–6.21), but important heterogeneity was observed
across studies (P < .001, I2 = 99.8%). For fiber-reinforced posts,
data on the incidence rate of root fractures were extractable from 5
RCTs (6, 21, 22, 25, 26) and 3 cohort studies (17, 23, 24). The
pooled estimate per 1000 post years was 4.78 (95% CI, 4.28–5.27).
Moreover, interstudy heterogeneity was high (P < .001, I2 = 100%)
(Fig. 2).
The incidence rate of root fractures for prefabricated metal posts
and carbon fiber posts was 2-fold higher compared with cast metal
posts and glass fiber posts (Figs. 2 and 3). However, subgroup
analysis failed to define the reasons for heterogeneity (Table 3). There
was no evidence of publication bias, as suggested by visual inspection of
the funnel plot and the Begg and Mazumbar rank correlation (P = .063
for metal-based posts, P = .107 for fiber-reinforced posts). The sensi-
tivity analysis showed that the pooled event rate for both systems did not
change substantially with the exclusion of any one study, thereby indi-
cating that the meta-analyses were not compromised in any way.
Incidence Rate of Noncatastrophic Failures
Data on the incidence rate of noncatastrophic failures for metal-
based posts were extractable from 4 RCTs (15, 16, 21, 26) and 3
cohort studies (19, 20, 27). Using a random-effects model (heteroge-
neity: P < .001, I2 = 100%), we found that the overall incidence rate of
noncatastrophic failures for metal-based posts was 12.69 (95% CI,
7.90–17.48) per 1000 post years. For fiber-reinforced posts, important
heterogeneity was also observed (P < .001, I2 = 100%), and the pooled
effect size was 19.39 (95% CI, 13.21–25.57) (Fig. 3).
The incidence rate of noncatastrophic failures was increased 2-
fold in cast metal posts and carbon fiber posts compared with prefab-
ricated metal posts and glass fiber posts, respectively. The reasons for
heterogeneity were also not defined by subgroup analysis (Table 3). The
robustness of the pooled effect size was examined by sequentially
removing each study and reanalyzing the remaining data sets. For

312 Figueiredo et al. JOE — Volume 41, Number 3, March 2015


Figure 2. A forest plot and funnel plot for the incidence rate of root fractures according to the post system.
metal-based posts, omitting the cohort study of Jung et al (20) resulted
in a significant reduction in the incidence rate of noncatastrophic fail-
ures (7.27; 95% CI, 3.36–11.18), whereas the other studies had no
substantial effect on the pooled results. However, there was no obvious
evidence of publication bias in this meta-analysis (Begg test, P = .548).
For fiber-reinforced posts, the results were not compromised by the
one-by-one study removals, and no obvious publication bias was found
(Begg test, P = .548) (Fig. 3).
Survival Rate of Post Systems
Figure 4 shows the study-specific and summary survival rates of the
post systems. The pooled survival rate for metal -based posts was 90.0%
(95% CI, 85.5–93.3), and statistically significant heterogeneity among
studies was observed (P = .008, I2 = 63%). For fiber-reinforced posts,
the pooled survival rate was 83.9% (95% CI, 67.6–92.8), and important
heterogeneity (P < .001, I2 = 97.3%) was also observed.
To explore the sources of heterogeneity, we considered perform-
ing a stratified analysis based on the follow-up periods according to the
Figure 3. A forest plot and funnel plot for the incidence rate of noncatastrophic failures according to the post system.
JOE — Volume 41, Number 3, March 2015
Review Article
subgroups of post systems (Table 4). The stratified analysis showed a
reduction in the overall survival rate of post systems over time. We
did not observe statistical heterogeneity between studies of prefabri-
cated metal posts, but significant changes in heterogeneity were noted
in the overall analysis of survival when cast metal posts were added. For
the fiber post system, heterogeneity was analyzed in 3 follow-up periods
(4.1–6 years, 6.1–8 years, and 8.1–10 years). Severe statistical hetero-
geneity was observed during the 6.1- to 8-year follow-up, but the vari-
ability in the individual study results was not related to the type of fiber
post. In addition, there was no evidence for a difference in the summary
estimates by adjusting the results for the study design (results not
shown). However, further analysis suggested that heterogeneity could
be primarily attributed to 2 studies (17,25) because of variations in
the sample sizes. After excluding these studies, the heterogeneity tests
showed no statistically significant differences (P = .547, I2 = 0.0%)
in the remaining studies.
Visual inspection of the funnel plot and the Begg and Mazumbar
rank correlation showed no potential for publication bias in any of
the meta-analyses. The ‘‘leave-one-out’’ method showed that the survival
Root Fracture of Post-retained Restorations 313
Review Article
TABLE 3. Subgroup Analysis to Investigate Differences between Studies Included in the Meta-analysis for Incidence Rate (/1000 post years) of Root Fracture and
Noncatastrophic Failures
Root fracture Noncatastrophic failures
Tests of Tests of
Pooled estimates heterogeneity Pooled estimates heterogeneity
No. of P value No. of P value
Source studies IR 95% CI (Q test) I2 (%) studies IR 95% CI (Q test) I2 (%)
Metal-based posts
Study design
RCT 4 5.01 2.28–7.74 <.001 99.7 4 6.93 3.11–10.75 <.001 99.9
Cohort 4 5.25 3.89–6.60 <.001 99.9 3 20.35 12.43–28.26 <.001 100
Sample size
<100 posts 5 4.66 2.94–6.35 <.001 99.6 5 14.60 4.11–25.09 <.001 100
$100 posts 3 5.92 4.37–7.48 <.001 99.9 2 7.93 2.33–13.52 <.001 100
Post-system
Prefabricated 8 5.52 4.38–6.66 <.001 99.9 7 12.80 9.13–16.48 <.001 100
metal post
Cast metal post 3 3.39 0.61 to 7.39 <.001 99.9 3 22.09 7.39–36.79 <.001 100
Fiber-reinforced posts
Study design
RCT 5 2.91 1.67–4.16 <.001 99.9 5 15.98 6.44–25.53 <.001 100
Cohort 3 8.19 3.20–13.18 <.001 100 3 25.08 6.89–43.27 <.001 100
Sample size
<100 posts 3 10.89 0.89–20.88 <.001 100 3 24.35 5.28–43.41 <.001 100
>100 posts 5 1.31 0.96–1.66 <.001 99.9 5 16.43 9.00–23.86 <.001 100
Post system
Glass fiber 3 3.58 0.62–6.54 <.001 100 3 14.17 4.86 to 33.20 <.001 100
Carbon fiber 5 5.69 3.46–7.91 <.001 100 5 22.56 15.04–30.08 <.001 100
CI, confidence interval; IR, incidence rate; RCT, randomized clinical trial.

rate for the adjusted and unadjusted data sets did not change substan-
tially with the exclusion of any one study. The results of the sensitivity
analysis exclude the possibility of bias and confirm the robustness of
effect sizes.
Discussion
The results of this systematic review show similarities between the
incidences of catastrophic failures among metal and fiber post–retained
restorations. There was a higher incidence of noncatastrophic failures
for fiber post–restored roots, but the survival rate was similar for both
types of posts. This systematic review included both clinical trials and
observational cohort studies, resulting in high levels of heterogeneity
among studies. Few studies evaluating metal and fiber posts were found
in the literature. Thus, considering the necessity of comparing these
types of posts, the 1-arm design for data analysis was performed.
Even the subgroup analyses did not identify the sources of heterogene-
ity, and the use of the 1-arm design for data analysis could explain the
high heterogeneity observed in the meta-analyses in the present study.
Interestingly, the survival rate analysis showed that the type of metal post
(cast or prefabricated) partially explained the interstudy heterogeneity
in this group. The same was not observed between the types of fiber

Figure 4. A forest plot and funnel plot for the survival rate of post systems.

314 Figueiredo et al. JOE — Volume 41, Number 3, March 2015

 Review Article
TABLE 4. Stratified Analysis Based on Periods of Follow-up According to the Post Systems*
Period of follow-up
Post-system/study 0–2 y 2.1–4 y 4.1–6 y 6.1–8 y 8.1–10 y 10.1–15 y
Metal post
Prefabricated metal post
Sterzenbach et al, 2012 — — 93.5 (81.7–97.9) — — —
Go
mez-Polo et al, 2010 — — — — 84.6 (66.9–94.9) —
Hikasa et al, 2010 99.3 (98.8–99.6) 97.0 (96.1–97.7) 94.2 (93.0–95.3) 91.4 (90.0–92.6) 86.6 (84.9–88.1) 78.7 (76.8–80.6)
Jung et al, 2007 — — — — 93.5 (92.3–94.6) —
Creugers et al, 2005 — — 96.0 (85.9–99.0) — — —
Ellner et al, 2003 — — — — 92.0 (67.5–99.6) —
King et al, 2003 100.0 (76.2–100.0) 100.0 (71.7–100.0) 100.0 (71.7–100.0) 89.0 (55.0–98.2) 89.0 (55.0–98.2) 89.0 (56.1–99.4)
Weine et al, 1991 — — — — 93.5 (88.0-96.6) —
Subgroup survival rate 99.3 (98.7–99.6) 97.0 (96.1–97.7) 94.2 (93.1–95.2) 91.4 (90.0–92.6) 87.0 (85.5–88.5) 78.7 (76.8–80.6)
Heterogeneity (Q test; I2) P = .217; 34.5% P = .814; 0.0% P = .939; 0.0% P = .778; 0.0% P = .893; 0.0% P = .417; 0.0%
Cast metal posts
Go
mez-Polo et al, 2010 — — — — 82.6 (73.5–89.5) —
Hikasa et al, 2010 97.0 (94.9–98.4) 93.7 (91.0–95.9) 89.6 (86.1–92.3) 82.5 (78.4–86.1) 75.5 (71.0–79.7) 55.4 (50.3–60.4)
Jung et al, 2007 — — — — 90.2 (78.1–96.8) —
Ellner et al, 2003 — — — — 100.0 (80.7–100.0) —
Subgroup survival rate 97.0 (94.9–98.4) 93.7 (91.0–95.9) 89.6 (86.1–92.3) 82.5 (78.4–86.1) 82.0 (72.5–88.7) 55.4 (50.3–60.4)
Heterogeneity (Q test; I2) NA NA NA NA P = .073; 56.9% NA
Overall survival rate 98.3 (94.2–99.5) 96.2 (95.3–97.0) 93.0 (89.4–95.4) 87.9 (77.7–93.7) 87.0 (81.6–91.0) 72.7 (50.2–87.6)
Heterogeneity (Q test; I2) P = .001; 84.8% P = .010; 78.4% P = .027; 63.6% P < .001; 92.2% P < .001; 77.9% P < .001; 97.6%
Fiber post
Carbon fiber
Ferrari et al, 2007 — — — 92.0 (90.1–93.6) — —
Segerstro € m et al, 2006 — — — 65.0 (55.1–73.7) — —
Mannocci et al, 2005 — — 90.0 (82.8–94.4) — — —
King et al, 2003 93.8 (72.8–99.7) 85.7 (60.3–97.5) 78.6 (52.1–94.2) 71.0 (45.4–87.8) 71.0 (45.4–87.8) —
Subgroup survival rate 93.8 (72.8–99.7) 85.7 (60.3–97.5) 88.1 (81.1–92.8) 79.5 (48.1–94.2) 71.0 (45.4–87.8) —
Heterogeneity (Q test; I2) NA NA P = .192; 41.2% P < .001; 96.7% NA NA
Glass fiber
Sterzenbach et al, 2012 — — 90.2 (77.5–96.1) — — —
Naumann et al, 2012 — — — — 46.0 (38.2–54.0) —
Ferrari et al, 2012 — — — 71.0 (64.9–76.4) — —
Signore et al, 2009 — — — 98.5 (97.0–99.3) — —
Subgroup survival rate NA NA 90.2 (78.1–96.8) 92.6 (33.2–99.7) 46.0 (38.1–54.7) NA
Heterogeneity (Q test; I2) NA NA NA P < .001; 98.6% NA NA
Overall survival rate 93.8 (72.8–99.7) 85.7 (60.3–97.5) 89.1 (83.2–93.0) 86.0 (67.6–94.7) 54.6 (31.0–76.2) NA
Heterogeneity (Q test; I2) NA NA P = .514; 0.0% P < .001; 97.3% P = .118; 49.2% NA

NA, not applicable.

*Data reported as survival rate followed by the 95% confidence interval in parentheses.

posts (carbon or glass fiber) when the survival rate was analyzed,
thereby stratifying the period of follow-up.
Despite the heterogeneity, meta-analyses of the outcomes were
performed. According to Dwyer et al (28), the decision to combine
the results of individual studies in the presence of heterogeneity de-
pends on the aim of the particular overview that is being undertaken.
If the aim is to identify the direction of a possible association, as per-
formed in the present study associating the type of posts and the inci-
dence of failures, then even with observational data, a meta-analysis
appears to be appropriate.
A recent meta-analysis of in vitro studies (3) evaluating the frac-
ture strength of roots restored with metal or fiber posts showed higher
values for metal posts, whereas the use of this last type of post resulted in
more catastrophic failures. Higher fracture strength can indicate longer
longevity under clinical function. The results of our meta-analysis of
clinical studies showed a higher survival rate for metal posts than for
fiber posts, primarily for longer periods of follow-up. However, the re-
sults of the present meta-analysis regarding the incidence rate of cata-
strophic failure contradict those observed for in vitro studies. In
contrast to the in vitro studies, a similarity in the incidence rate of cata-
strophic failures was observed among metal and fiber posts. Interest-
ingly, when only clinical trials are analyzed, metal posts showed a
higher incidence rate of catastrophic failures. The opposite was
observed for cohort studies. Another interesting observation on sub-
group analysis was the tendency to reduce catastrophic failures and in-
crease noncatastrophic failures when cast posts (for metal based) and
glass fiber posts (for fiber reinforced) were used. Regardless of the CI,
no significant difference was observed.
Only studies using follow-up periods longer than 5 years were
included in the present meta-analysis. A recent recommendation for
conducting controlled trials on dental materials (29) advised an obser-
vational period of longer than 5 years for indirect restorations in clinical
studies. All studies included in this systematic review used indirect
restoration to restore the tooth that received the post, requiring
follow-up periods longer than 5 years to assess the success of the restor-
ative procedure. Furthermore, some clinical studies that were excluded
because of shorter follow-up periods were older publications of
included studies that reported on the same patients.
Some limitations should be acknowledged in this meta-analysis.
First, almost all of the observational studies that met the inclusion criteria
were retrospective cohort studies, which have less control over the sub-
ject selection and measurements and the risk for confounding. However,
cohort studies often include more subjects and have a longer follow-up
time. In addition, in regard to assessing harm, such as the incidence of
failures, these types of studies are a good source of data. Second, studies
that suffer from a high risk of bias were not excluded from the meta-
analysis. This decision was made because of the lack of unbiased studies
that met the inclusion criteria for this review. Finally, we were unable to
explore the risk factors for the outcomes of interest, such as the amounts
of coronal remaining, the presence of a ferule effect, the location of the
tooth, and occlusion features. Unfortunately, these risk factors were not
described in most of the included studies.

JOE — Volume 41, Number 3, March 2015 Root Fracture of Post-retained Restorations 315
Review Article
In conclusion, the results of the present systematic review based
on clinical studies do not support the indication of fiber-reinforced
posts based on a reduction of catastrophic failures. However, this review
also showed the need for further well-designed clinical studies evalu-
ating intraradicular posts.
Acknowledgments
The authors deny any conflicts of interest related to this study.
References
1. Cheung W. A review of the management of endodontically treated teeth. Post, core
and the final restoration. J Am Dent Assoc 2005;136:611–9.
2. Bateman G, Ricketts DNJ, Saunders WP. Fibre-based post systems: a review. Br Dent
J 2003;195:43–8. discussion 37.
3. Zhou L, Wang Q. Comparison of fracture resistance between cast posts and fiber
posts: a meta-analysis of literature. J Endod 2013;39:11–5.
4. Santos AF, Meira JB, Tanaka CB, et al. Can fiber posts increase root stresses and
reduce fracture? J Dent Res 2010;89:587–91.
5. Zicari F, Van Meerbeek B, Debels E, et al. An up to 3-year controlled clinical trial
comparing the outcome of glass fiber posts and composite cores with gold alloy-
based posts and cores for the restoration of endodontically treated teeth. Int J Pros-
thodont 2011;24:363–72.
6. Ferrari M, Vichi A, Fadda GM, et al. A randomized controlled trial of endodontically
treated and restored premolars. J Dent Res 2012;91:72S–8.
7. Higgins JPT, Altman DG, Gøtzsche PC, et al. The Cochrane Collaboration’s tool for
assessing risk of bias in randomised trials. BMJ 2011;343:d5928.
8. Wells GA, Shea B, O’Connell D, et al. The Newcastle-Ottawa Scale (NOS) for As-
sessing the Quality of Nonrandomised Studies in Meta-analyses. Canada: Depart-
ment of Epidemiology and Community Medicine, University of Ottawa; 2000.
9. Bhaumik DK, Amatya A, Normand S-L, et al. Meta-analysis of rare binary adverse
event data. J Am Stat Assoc 2012;107:555–67.
10. Cochran WG. The combination of estimates from different experiments. Biometrics
1954;10:101.
11. Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med
2002;21:1539–58.
12. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. A basic introduction to fixed-
effect and random-effects models for meta-analysis. Res Synth Methods 2010;1:
97–111.
13. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for pub-
lication bias. Biometrics 1994;50:1088–101.
316 Figueiredo et al.
14. Sterne JAC, Egger M, Smith GD. Investigating and dealing with publication and other
biases in meta-analysis. BMJ 2001;323:101–5.
15. Creugers NHJ, Kreulen CM, Fokkinga WA. A 5-year prospective clinical study on core
restorations without covering crowns. Int J Prosthodont 2005;18:40–1.
16. Ellner S, Bergendal T, Bergman B. Four post-and-core combinations as abutments
for fixed single crowns: a prospective up to 10-year study. Int J Prosthodont 2003;
16:249–54.
17. Ferrari M, Cagidiaco MC, Goracci C, et al. Long-term retrospective study of the clin-
ical performance of fiber posts. Am J Dent 2007;20:287–91.
18. G
omez-Polo M, Llid
o B, Rivero A, et al. A 10-year retrospective study of the survival
rate of teeth restored with metal prefabricated posts versus cast metal posts and
cores. J Dent 2010;38:916–20.
19. Hikasa T, Matsuka Y, Mine A, et al. A 15-year clinical comparative study of the cu-
mulative survival rate of cast metal core and resin core restorations luted with ad-
hesive resin cement. Int J Prosthodont 2010;23:397–405.
20. Jung RE, Kalkstein O, Sailer I, et al. A comparison of composite post buildups and
cast gold post-and-core buildups for the restoration of nonvital teeth after 5 to 10
years. Int J Prosthodont 2007;20:63–9.
21. King PA, Setchell DJ, Rees JS. Clinical evaluation of a carbon fibre reinforced carbon
endodontic post. J Oral Rehabil 2003;30:785–9.
22. Mannocci F, Qualtrough AJE, Worthington HV, et al. Randomized clinical compar-
ison of endodontically treated teeth restored with amalgam or with fiber posts and
resin composite: five-year results. Oper Dent 2005;30:9–15.
23. Naumann M, Koelpin M, Beuer F, Meyer-Lueckel H. 10-year survival evaluation for
glass-fiber-supported postendodontic restoration: a prospective observational clin-
ical study. J Endod 2012;38:432–5.
24. Segerstr€om S, Astb€ack J, Ekstrand KD. A retrospective long term study of teeth
restored with prefabricated carbon fiber reinforced epoxy resin posts. Swed Dent
J 2006;30:1–8.
25. Signore A, Benedicenti S, Kaitsas V, et al. Long-term survival of endodontically
treated, maxillary anterior teeth restored with either tapered or parallel-sided
glass-fiber posts and full-ceramic crown coverage. J Dent 2009;37:115–21.
26. Sterzenbach G, Franke A, Naumann M. Rigid versus flexible dentine-like endodontic
posts—clinical testing of a biomechanical concept: seven-year results of a random-
ized controlled clinical pilot trial on endodontically treated abutment teeth with se-
vere hard tissue loss. J Endod 2012;38:1557–63.
27. Weine FS, Wax AH, Wenckus CS. Retrospective study of tapered, smooth post systems
in place for 10 years or more. J Endod 1991;17:293–7.
28. Dwyer T, Couper D, Walter SD. Sources of heterogeneity in the meta-analysis of
observational studies: the example of SIDS and sleeping position. J Clin
Epidemiol 2001;54:440–7.
29. Hickel R, Roulet J-F, Bayne S, et al. Recommendations for conducting controlled
clinical studies of dental restorative materials. Clin Oral Investig 2007;11:5–33.
JOE — Volume 41, Number 3, March 2015