Diagnostic accuracy of CBCT for tooth fractures: A meta-analysis

$Page 1: Diagnostic accuracy of CBCT for tooth fractures: A meta-analysis$
JJOD-2198; No. of Pages 9

Review

Diagnostic accuracy of CBCT for tooth fractures:A meta-analysis

Hu Long, Yang Zhou, Niansong Ye, Lina Liao, Fan Jian,Yan Wang, Wenli Lai *

Department of Orthodontics, State Key Laboratory of Oral Diseases, West China Hospital of Stomatology, Sichuan

University, Chengdu 610041, Sichuan, China

2

j o u r n a l o f d e n t i s t r y x x x ( 2 0 1 3 ) x x x – x x x

a r t i c l e i n f o

Article history:

Received 17 July 2013

Received in revised form

26 November 2013

Accepted 30 November 2013

Available online xxx

Keywords:

Cone-beam computed tomography

Diagnostic accuracy

Meta-analysis

Root fracture

Systematic review

Tooth fracture

a b s t r a c t

Objectives: The objective of this meta-analysis was to determine the diagnostic accuracy of

cone-beam computed tomography (CBCT) for tooth fractures in vivo.

Methods: PubMed, Embase, Web of Science, ProQuest Dissertations & Theses, CNKI and

SIGLE were searched from January 1990 to April 2013 for eligible studies. Quality Assessment

of Diagnostic Accuracy Studies-2 (QUADAS-2) was employed to assess the quality of the

included studies. Meta-analyses were performed in MetaDisc 1.4, Stata 12.1 and StatsDirect

2.7.9.

Results: Finally, 12 studies were included in this meta-analysis. The pooled sensitivity,

specificity, positive likelihood ratio, negative likelihood ratio and summary receiver oper-

ating characteristic were 0.92 (95% CI = 0.89–0.94), 0.85 (95% CI = 0.75–0.92), 5.68 (95%

CI = 3.42–9.45), 0.13 (95% CI = 0.09–0.18) and 0.94 (95% CI = 0.90–0.98), respectively. The

pooled prevalence of tooth fractures in patients with clinically-suspected but periapical-

radiography-undetected tooth fractures was 91% (95% CI = 83%-97%). Positive and negative

predictive values were 0.98 and 0.43 (subgroup analysis: 0.98 and 0.28 for endodontically-

treated teeth; 0.99 and 0.77 for non-endodontically-treated teeth).

Conclusion: We suggest that CBCT has a high diagnostic accuracy for tooth fractures and

could be used in clinical settings. We can be very confident with positive test results but

should be very cautious with negative test results, especially for endodontically treated

teeth.

# 2013 Elsevier Ltd. All rights reserved.

Available online at www.sciencedirect.com

ScienceDirect

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

1. Introduction

With the evolution and adaptation of animal species to diet,

their teeth have evolved sophisticated microstructures to

absorb considerably high occlusal forces and to protect their

teeth from fractures,1 although their dental enamel is per se

* Corresponding author at: Department of Orthodontics, State Key LaSichuan University, No. 14, Section 3, Ren Min South Road, Chengdu

E-mail address: [email protected] (W. Lai).

Please cite this article in press as: Long H, et al. Diagnostic accuracy of Chttp://dx.doi.org/10.1016/j.jdent.2013.11.024

0300-5712/$ – see front matter # 2013 Elsevier Ltd. All rights reservedhttp://dx.doi.org/10.1016/j.jdent.2013.11.024

highly brittle. Despite these sophisticated microstructures,

tooth fractures are still frequently encountered in dental

practice.3 It has been well documented that the most common

cause of tooth fractures in adults is high occlusal forces caused

by biting on hard objects or abnormal contacts of opposing

teeth.4 In addition, earlier endodontic therapy and malocclu-

sions are implicated in the aetiology of tooth fractures.2,3,5 The

boratory of Oral Diseases, West China Hospital of Stomatology,610041, China. Tel.: +86 2885501442.

BCT for tooth fractures: A meta-analysis. Journal of Dentistry (2013),

.

http://dx.doi.org/10.1016/j.jdent.2013.11.024

mailto:[email protected]


http://www.sciencedirect.com/science/journal/03005712

http://dx.doi.org/www.intl.elsevierhealth.com/journals/jden


j o u r n a l o f d e n t i s t r y x x x ( 2 0 1 3 ) x x x – x x x2


treatment of tooth fractures often requires an interdisciplin-

ary approach, e.g., combined endodontic–orthodontic–

prosthodontic treatments.6 Thus, tooth fracture is an issue

discussed in almost all the dental specialties.

Unfortunately, the diagnosis of tooth fractures, especial-

ly root fractures, is sometimes difficult for dental practi-

tioners due to their extremely variable clinical presentations

and no pathognomonic signs.7 The most common diagnos-

tic modality for tooth fractures—periapical radiography—

has been revealed to have a low diagnostic accuracy and to

leave many tooth fractures undetected.8–10 The introduction

of cone-beam computed tomography (CBCT) in dentistry

has allowed dental practitioners to acquire three-dimen-

sional visualization of teeth with high spatial resolution and

low radiation.11 Several studies have revealed the outstand-

ing diagnostic accuracy of CBCT for tooth fractures,12–15

especially for those undetected through periapical radiog-

raphy.16 However, these studies based on in vitro tooth

fractures and the diagnostic performance of CBCT for tooth

fractures in vivo has been poorly understood.9,10 Moreover, it

has been reported that not all dental practitioners or dental

students were aware of CBCT.17,18 Therefore, in order to

help practitioners make correct diagnosis and subsequent

prudent treatment decisions, we conducted this meta-

analysis to critically appraise the currently available evi-

dence and to determine the diagnostic accuracy of CBCT for

tooth fractures in vivo.

2. Methods

2.1. Inclusion criteria for included studies

Studies should be those which examined the diagnostic

accuracy of CBCT for tooth fractures in vivo. In vitro studies

and case reports would be excluded. Participants in each study

should be clinically suspected with tooth fractures and

received CBCT examinations for the diagnosis of tooth

fractures. The number of participants in each study should

be at least 10. Furthermore, studies should have reference

tests—surgical exploration or extractions—to establish the

diagnosis of tooth fractures.

Table 1 – Search strategies for each database.

Database

PubMed (Cone-Beam Computed Tomography [Mes

OR CBCT) AND (Tooth Fractures [Mesh] O

Embase (Cone-Beam Computed Tomography OR c

CBCT) AND (tooth fracture OR tooth fract

fractures)

Web of Science (Cone-Beam Computed Tomography OR c

CBCT) AND (tooth fracture OR tooth fract

fractures)

ProQuest Dissertation &

Theses

(cone-beam computed tomography OR con

fracture)

CNKIa (cone beam CT OR CBCT) AND (tooth frac

SIGLE CBCT AND tooth fracture

a Chinese terms were used in CNKI, its search strategy listed above wer

April 2013.


2.2. Search methods

We searched the electronic databases of PubMed, Embase,

Web of Science, ProQuest Dissertations & Theses and CNKI.

Moreover, the grey literature database of SIGLE was searched.

The specific search strategy for each database is presented in

Table 1. The electronic searching was from January 1990 to

April 2013 with no language restriction. Two review authors

(HL and YZ) conducted the electronic searching independently

and in duplicate. Any disagreement was solved by discussion

or referred to a third author.

2.3. Data extraction and analysis

2.3.1. Data extraction

Demographic data, reference test, true positive, false positive,

false negative and true negative were extracted and recorded

independently and in duplicate by two reviewer authors (HL

and YZ).

2.3.2. Study outcomesStudy outcomes were sensitivity, specificity, positive likeli-

hood ratio (LR), negative LR and summary receiver operating

characteristic (SROC). Moreover, derivative outcomes (preva-

lence, positive predictive value (PPV) and negative predictive

value (NPV)) were calculated from the study outcomes

mentioned above.

2.3.3. Quality assessmentThe quality of each included study was evaluated according to

Quality Assessment of Studies of Diagnostic Accuracy-2

(QUADAS-2).19 Specifically, for each category (risk of bias

and applicability concerns), studies with two or more domains

of high risk would be designated as high risk; those with only

one domain of high risk would be designated as medium risk;

those with no domain of high risk would be designated as low

risk. Assessments were conducted independently and in

duplicate by two review authors.

2.3.4. Data analysisOutcome data, i.e., sensitivity, specificity, positive LR, negative

LR and SROC, would be statistically pooled. Moreover,

Search strategy

h] OR cone-beam CT OR cone beam CT OR cone-beam OR cone beam

R tooth fracture* OR dental fracture* OR root fracture*)

one-beam CT OR cone beam CT OR cone-beam OR cone beam OR

ures OR dental fracture OR dental fractures OR root fracture OR root

one-beam CT OR cone beam CT OR cone-beam OR cone beam OR

ures OR dental fracture OR dental fractures OR root fracture OR root

e beam CT OR CBCT) AND (tooth fracture OR dental fracture OR root

ture OR root fracture OR fracture)

e English translations. Limits: publication date from January 1980 to



Fig. 1 – PRISMA flow diagram for studies retrieved through

the searching and selection processes.

j o u r n a l o f d e n t i s t r y x x x ( 2 0 1 3 ) x x x – x x x 3


prevalences of tooth fractures, non-fractures, positive test

results and negative test results were statistically pooled

through Freeman–Tukey arcsine square root transformation

for variance stabilization.20,21 PPV and NPV were calculated

through Bayes’ theorem.22 Heterogeneity across studies was

assessed through I2 statistic. An I2 statistic greater than 50%

was considered substantial heterogeneity. If substantial

heterogeneity existed, subgroup analysis or meta-regression

would be conducted to explore the heterogeneity. Mantel–

Haenszel fixed effect model would be used when a common

effect size is shared among includes studies; otherwise,

DerSimonian–Laird random effect model would be used.23

In addition, even if a common effect size is shared among all

studies, a random effect model would be adopted when

significant heterogeneity existed (I2 > 50%).23

Sensitivity analysis was performed to assess the robust-

ness of the pooled results in the meta-analyses. Moreover,

publication bias was evaluated based on a regression test and

funnel plot designed specifically for diagnostic systematic

reviews.24 All the statistical analyses were performed in Meta-

Disc 1.4,25 Stata 12.1 (StataCorp, College Station, TX, USA) and

StatsDirect 2.7.9 (StatsDirect Ltd., Cheshire, UK).

3. Results

3.1. Description of included studies

Initially, we identified 154 articles from the database and

excluded 137 irrelevant articles. The remaining 17 articles

were further assessed for eligibility and 128–10,26–34 were finally

included in this meta-analysis. Specifically, the detailed

procedures of electronic searching are shown in Fig. 1. Among

Table 2 – Characteristics of included studies.

Study Samplesize (teeth)

Age(yrs)

Fracturetype

Tooth statu

Du (2010) 40 36.6 Vertical Non-endodont

Liu (2010) 40 57 Vertical and

oblique

–

Edlund (2011) 32 20–70 Vertical Endodontic

Ning (2011) 17 63 Vertical Both

Qin (2011) 47 26–58 – –

Wang (2011) 135 22–82 Vertical and

oblique

Both

Xue (2011) 84 20–89 Vertical Both

Yan (2011) 23 – Vertical Both

Ding (2012) 38 57 Alla Both

Kajan (2012) 10 21–70 All Endodontic

Metska (2012) 39 – Vertical Endodontic

Liu (2013) 31 45 – –

a All is indicative of vertical, oblique and horizontal fractures.


the included studies, sample sizes ranged from 10 to 135 teeth.

Fracture types included vertical, oblique and horizontal tooth

fractures. Both non-endodontically and endodontically trea-

ted teeth were included. All studies verified tooth fractures

through reference standards, i.e., surgical extraction or

exploration. The detailed characteristics of each study are

presented in Table 2.

3.2. Quality assessment

According to QUADAS-2, the quality assessment comprised two

categories: risk of bias and applicability concerns. In this meta-

analysis, although risk of bias of included studies was medium

to high, the applicability was of medium to low risk (Table 3).

s CBCT Reference test

ic NewTom 3 G; 110 kV, 6 mA, 9 s Surgical extraction

or exploration

Planmeca ProMax 3D Surgical extraction

or exploration

iCAT or 3D Accuitomo 80;

voxel: 80 or 125 mm

Surgical exploration

3DX multi-image micro-CT; 80 kV,

5 mA, 20 s; voxel: 125 mm

Surgical extraction

or exploration

Kodak 9000C 3D Surgical extraction

or exploration

3DX Accuitomo; 80 kV, 5 mA,

17.5 s; voxel: 125 mm

Surgical extraction

or exploration

Galileos Comfort; 85 kV,

35 mA s; voxel: 125 mm

Surgical extraction

or exploration

3DX multi-image micro-CT;

80 kV, 5 mA, 17.5 s

Surgical extraction

KaVo 3D eXam; 90–120 kV,

3–7 mA; voxel 125 mm

Surgical extraction

or exploration

NewTom VG; 110 kV, 5.5 mA, 5.4 s Surgical extraction

NewTom 3 G or 3D Accuitomo 170;

110 kV, 3.9–5.6 mA or 90 kV, 5 mA;

voxel: 200 or 80 mm

Orthograde retreatment,

surgical exploration

or extraction

Planmeca ProMax 3D Surgical extraction



Table 3 – Quality assessment of the included studies according to QUADAS-2.

Study Risk of bias Applicability concerns

Patientselection

Indextest

Reference Flow &timing

Total Patientselection

Indextest

Reference Total

Du (2010) High Low High High High Low Low High Medium

Liu (2010) High Unclear Low Low Medium Low Unclear Low Low

Edlund (2011) High High Low Low High Low High Low Medium

Ning (2011) High Low Low Low Medium Low Low Low Low

Qin (2011) High Low High High High Low Low High Medium

Wang (2011) High Low Low Low Medium Low Low Low Low

Xue (2011) Unclear Low High High High Unclear Low High Medium

Yan (2011) High Unclear High High High Low Unclear High Medium

Ding (2012) High Unclear Low Low Medium High Unclear Low Medium

Kajan (2012) High Unclear Low High High Low Unclear Low Low

Metska (2012) High Low Low High High Low Low Low Low

Liu (2013) Unclear Unclear Low Low Low Unclear Unclear Low Low

High: high risk; unclear: unclear risk; low: low risk.



3.3. Description of outcomes

All the study outcomes proposed above were evaluated among

the included studies, i.e., sensitivity, specificity, positive LR,

negative LR and SROC.

3.4. Statistical pooling of outcomes

3.4.1. Sensitivity and specificityConsidering that a common effect size was shared among

studies and no significant heterogeneity existed for

Fig. 2 – The pooled diagnostic indices for the diagnosis of tooth

diagnosis of tooth fractures through CBCT; (B) the pooled specif

the pooled positive likelihood for the diagnosis of tooth fracture

the diagnosis of tooth fractures through CBCT.


sensitivity and specificity (I2 = 38.7% and 24.5%, respective-

ly), the Mantel–Haenszel fixed effect model was employed.

As displayed in Fig. 2 A and B, the results showed

that the pooled sensitivity and specificity were 0.92

(95% CI = 0.89–0.94) and 0.85 (95% CI = 0.75–0.92), respective-

ly.

3.4.2. Positive LR and negative LR

Since a common effect size was shared among studies and no

significant heterogeneity existed (I2 = 26.7% and 0.0%, respec-

tively), fixed effect model was used. The results revealed that

fractures through CBCT. (A) The pooled sensitivity for the

icity for the diagnosis of tooth fractures through CBCT; (C)

s through CBCT; (D), the pooled negative likelihood ratio for



Fig. 3 – Summary receiver operating characteristics (SROC) curve for the diagnosis of tooth fractures through CBCT.



the pooled positive LR and negative LR were 5.68 (95%

CI = 3.42–9.45) and 0.13 (95% CI = 0.09–0.18), respectively

(Fig. 2C and D).

3.4.3. SROCThe SROC curve for discrimination of fractures and non-

fractures is presented in Fig. 3. The pooled area under curve

(AUC) of SROC was 0.94 (95% CI = 0.90–0.98).

Fig. 4 – The pooled prevalence of tooth fractures in patients

with clinically-suspected but periapical-radiography-

undetected tooth fractures.


3.5. Prevalences

Although a common effect size was shared among the

included studies, random effect model was employed due to

significant heterogeneity (I2 = 87.6% for tooth fractures and

non-fractures; I2 = 87.2% for positive and negative test results).

The meta-analysis revealed that the pooled prevalence of

tooth fractures was 0.91 (95% CI = 0.83–0.97) (Fig. 4). Moreover,

the results showed that the pooled prevalences of non-

fractures, positive test results and negative test results were

0.09 (95% CI = 0.03–0.17), 0.87 (95% CI = 0.78–0.94) and 0.13 (95%

CI = 0.06–0.22), respectively.

3.6. PPV and NPV

Through Bayes’ theorem,22 PPV and NPV were calculated to be

0.98 and 0.43. Moreover, PPV and NPV were 0.99 and 0.77 for

non-endodontically treated teeth; PPV and NPV were 0.98 and

0.28 for endodontically treated teeth.

3.7. Sensitivity analysis

Since reference standard test was not applied in all the

patients in three studies,26,29,32 a sensitivity analysis was

performed by excluding these three studies, but no significant

changes were found (Table 4).

Seven included studies exhibited high risk of bias (Table 3),

thus we performed a sensitivity analysis by excluding these

studies with high risk of bias and found a moderate but non-

significant increase in diagnostic accuracy (Table 4).

Changes in effect models (fixed vs. random effect model)

failed to reveal any significant change in the pooled results

(Table 4).



Table 4 – Sensitivity analysis and subgroup analysis.

Item Sensitivity Specificity Positive LR Negative LR SROC PPV NPV

Original estimate 0.92 (0.89, 0.94) 0.85 (0.75, 0.92) 5.68 (3.42, 9.45) 0.13 (0.09, 0.18) 0.94 (0.90, 0.98) 0.98 0.43

Random effect model 0.92 (0.89, 0.94) 0.85 (0.75, 0.92) 3.49 (1.91, 6.40) 0.13 (0.09, 0.18) 0.94 (0.90, 0.98) 0.97 0.43

Gold standard reference test 0.91 (0.88, 0.94) 0.85 (0.75, 0.92) 5.76 (3.35, 9.91) 0.13 (0.09, 0.20) 0.94 (0.89, 0.99) 0.98 0.43

Exclusion of high risk of bias 0.94 (0.90, 0.97) 0.93 (0.81, 0.99) 10.98 (4.12, 29.29) 0.10 (0.06, 0.18) 0.98 (0.95, 1.00) 0.99 0.50

Inclusion of only low risk of

applicability

0.92 (0.88, 0.95) 0.86 (0.75, 0.94) 6.35 (3.35, 12.05) 0.13 (0.08, 0.21) 0.95 (0.89, 1.00) 0.98 0.43

Subgroup analysis on

vertical fracture

0.89 (0.84, 0.93) 0.73 (0.55, 0.87) 3.40 (1.87, 6.17) 0.18 (0.10, 0.29) 0.88 (0.77, 0.99) 0.97 0.35


endodontic teeth

0.82 (0.72, 0.89) 0.82 (0.68, 0.92) 4.19 (2.82, 7.69) 0.25 (0.16, 0.39) 0.90 (0.81, 0.98) 0.98 0.28


non-endodontic teeth

0.97 (0.92, 0.99) 0.95 (0.76, 1.00) 13.70 (3.01, 62.42) 0.03 (0.01, 0.11) – 0.99 0.77

PPV: positive predictive value; NPV: negative predictive value.



Among the included studies, fracture types consisted of

vertical, oblique and horizontal (Table 2). However, the

differential diagnostic accuracy of CBCT can only be assessed

for vertical fracture. Thus, we conducted a subgroup analysis

on vertical tooth fracture and found that a moderate but non-

significant decrease in diagnostic accuracy (Table 4).

Both endodontically and non-endodontically treated teeth

were studied in the included studies. Among the included

studies, only Wang 201130 evaluated the differential diagnostic

accuracy of CBCT for each type. In total, two and four studies

assessed the differential diagnostic accuracy of CBCT for non-

endodontically and endodontically treated teeth, respectively.

Thus, we performed a subgroup analysis and found that the

sensitivity of CBCT was significantly higher for non-endodon-

tically treated teeth (sensitivity: 0.97, 95% CI = 0.92–0.99) than

for endodontically treated teeth (sensitivity: 0.82, 95% CI = 0.72–

0.89). Moreover, a moderate but non-significant increased

diagnostic accuracy was found for non-endodontically treated

teeth regarding specificity, positive LR and negative LR (Table 4).

3.8. Cumulative meta-analysis

Cumulative met-analysis was performed to determine the

chronological changes in diagnostic accuracy of CBCT for

tooth fractures. The results showed no significant changes

over time (data not shown).

3.9. Assessment of publication bias

Since Egger’s regression test and Begg’s rank correlation test

would inflate Type I error and are unsuitable for diagnostic

systematic review, we employed an adapted Egger’s and

Begg’s tests specifically for diagnostic systematic reviews.24

The results revealed that neither adapted Egger’s (p = 0.12) nor

Begg’s test (p = 0.84) indicated any evidence of publication

bias.

4. Discussion

In this meta-analysis, the included 12 studies evaluated five

outcomes. The pooled sensitivity, specificity, positive LR,

negative LR and SROC were 0.92, 0.85, 5.68, 0.13 and 0.94,

respectively. Sensitivity analysis and subgroup analysis


showed consistent results in the meta-analysis except for

the subgroup analysis on non-endodontically and endodonti-

cally treated teeth. Cumulative meta-analysis revealed no

significant changes over time regarding the diagnostic

accuracy of CBCT for tooth fractures. Furthermore, neither

modified Egger’s test nor Begg’s test was indicative of any

evidence of publication bias. Thus, in general, the results in

the meta-analysis were robust.

According to QUADAS-2,19 qualities of included studies for

diagnostic systematic review comprised two categories: risk of

bias and applicability concerns. Specifically, risk of bias

evaluates the potential for bias in each included studies while

applicability concerns assess the potential for generalization

or applicability of the results to real clinical scenarios. The

results showed that risk of bias was medium to high and the

applicability was of medium to low risk. In almost all the

included studies, prior to CBCT exams, subjects were firstly

diagnosed through periapical radiography but no convincing

diagnosis was established. With regards to risk of bias, these

included studies suffered from a significant bias—inappropri-

ate patient selection. Specifically, these periapical-radiogra-

phy-undetected patients were ‘‘difficult-to-diagnose’’ patients

and may result in an underestimation of diagnostic accuracy

of CBCT in this meta-analysis. Thus, high risk of bias was

assigned to the category—risk of bias. Despite high risk of bias,

the sensitivity analysis on exclusion of high risk of bias

resulted in no significant changes in the results (Table 4),

indicative of robustness of the results from the meta-analysis.

Moreover, this is closer to real clinical scenarios: patients with

clinically-suspected tooth fractures would be first diagnosed

through periapical radiographs and, if undetected, would be

further diagnosed by using CBCT. Thus, low risk was assigned

to the category—applicability concerns. Similarly, a sensitivity

analysis on inclusion of studies with low risk of applicability

failed to reveal any significant change (Table 4). Since CBCT for

tooth fracture is a clinical application rather than a theoretical

issue, we would focus on its applicability and suggest that

quality of the included studies in this meta-analysis are

medium to high (medium to low risk).

In this meta-analysis, all included studies employed

surgical extraction or exploration to establish the reference

standard results. However, among the included studies,

three26,29,32 did not apply the reference standard tests in all

the patients, which may bias the results in the meta-analysis.





We performed a sensitivity analysis by excluding the three

studies and found no significant changes from the original

estimates, which indicates that the results in the meta-

analysis were robust and not biased by these three studies.

The results revealed that pooled sensitivity and specificity

were 0.92 and 0.85, respectively. By definition, sensitivity is the

proportion of those with disease who have positive test results

while specificity is the proportion of those without diseases who

have negative test results. Thus, the results indicated that,

among a group of patients with both tooth fractures and no

fractures, 92% of the tooth fractures could be detected and 85%

of the non-fractures could be ruled out through CBCT. Or

alternatively, for a patient with tooth fractures, the chance of

the diagnosis of tooth fracture is 92%; for a healthy person

without tooth fracture, the chance of the diagnosis of non-

fracture is 85%.

However, there would always be a trade-off between

sensitivity and specificity.35 Thus, the two indices cannot

reflect the diagnostic accuracy well individually. For this

purpose, SROC has been proposed as an index to assess the

diagnostic accuracy in meta-analyses36–38 and has been used in

many recent diagnostic meta-analyses.35,39 SROC curve was

constructed by plotting sensitivity against 1-specificity, thereby

integrating both sensitivity and specificity into one index. In

practice, the area under curve (AUC) of SROC has a range from

0.5 (no better than chance) to 1 (perfect test).38 In our study, the

meta-analysis revealed that the pooled AUC of SROC was 0.94

(95% CI = 0.90–0.98), which indicates that CBCT has a high

diagnostic accuracy for the diagnosis of tooth fractures.

The meta-analysis showed that the pooled positive LR and

negative LR were 5.68 and 0.13, respectively. By definition,

positive LR is the ratio of the true positive rate to the false

positive rate and negative LR is the ratio of the false negative

rate to the true negative rate. Thus, a diagnosis would be more

accurate when positive LR becomes higher and negative LR

becomes lower. It has been documented that positive LR

greater than 5 or negative LR less than 0.2 can provide strong

diagnostic evidence.40 Therefore, our results (positive

LR = 5.68 > 5; negative LR = 0.13 < 0.2) indicate that CBCT is

able to give a strong diagnostic value regarding tooth fractures.

Given the prevalence of tooth fractures, through Bayes’

theorem,22 we are able to calculate the probabilities of true

fracture and non-fracture among positive and negative test

results, defined as positive and negative predictive values,

respectively. The specific patient population in this meta-

analysis is those with clinically suspected tooth fractures but

undetected through periapical radiographs. The meta-analy-

sis showed that the pooled prevalence of tooth fracture in this

specific patient population was 0.91 (95% CI = 0.83–0.97). Then,

the positive and negative predictive values were calculated to

be 0.98 and 0.43. This indicates that, for a given patient

diagnosed with tooth fractures by CBCT in this specific patient

population, we have 98% confidence to say that he really has

tooth fracture. Moreover, for a given patient diagnosed with

non-fractures by CBCT, we have only 43% confidence to say

that he is really free from tooth fractures, even worse than

chance. However, these positive and negative predictive

values would be susceptible to changes for different patient

populations with different prevalences of tooth fractures41:

the positive and negative predictive values would be 0.85 and


0.88 if the prevalence of tooth fracture is 50%. Thus, the

positive and negative predictive values calculated in this

meta-analysis could be used in only the aforementioned

specific patient population—those with clinically suspected

tooth fractures but undetected by periapical radiographs. In

real clinical scenarios, if a patient (clinically suspected tooth

fractures but undetected by periapical radiographs) is diag-

nosed with tooth fractures through CBCT, we can be very

confident that he suffered from tooth fractures and refer him

to have his fractured teeth treated. If a patient is diagnosed

without tooth fractures through CBCT (chances of negatives

are very low in the real clinical scenarios since the pooled

prevalence of negative test result is 13% for this subgroup of

patients), we have only 43% of confidence of non-fractures and

we suggest that a close follow-up is needed.

Tooth fractures can be divided into different types based on

the directions of fractures: vertical, horizontal, oblique. For

different types of tooth fractures, diagnoses may be different.

Unfortunately, in this meta-analysis, due to unavailability of

original data, we were able to perform a subgroup analysis

only for vertical tooth fractures, but failed to find any

significant change from the original estimates (Table 4). This

finding suggests the diagnostic accuracy of CBCT is similar

among different types of tooth fractures.

It has been reported that the diagnostic accuracy of CBCT

for endodontically-treated teeth can be reduced,42,43 which

can be attributed to potential artefacts caused by root canal

fillings. Consistently, in this meta-analysis, we found that

sensitivity was significantly higher in non-endodontically

treated teeth than in endodontically treated teeth [0.97 (95%

CI = 0.92–0.99) vs. 0.82 (95% CI = 0.72–0.89)] while other indices

(specificity, positive LR, negative LR and SROC) were similar.

This indicates that, in clinical scenarios, the chance of

diagnosis of tooth fractures is 97% for a non-endodontically

treated tooth with fractures and that chance would be 82% for

an endodontically treated tooth with fractures. Moreover, let

the prevalence be 91%, the positive predictive values were

calculated to be similar: 0.98 and 0.99 for endodontically and

non-endodontically treated teeth, respectively; while negative

predictive values were 0.28 and 0.77 for endodontically and

non-endodontically treated teeth, respectively. Based on the

CBCT results, we can easily know whether a tooth is an

endodontically or non-endodontically treated tooth. If the

result is positive, regardless of whether it is an endodontically

or non-endodontically treated tooth, it is almost 100% chance

of tooth fractures. However, if the result is negative, chances

of false negatives are much higher for endodontically treated

teeth (100–28% = 72%) than for non-endodontically treated

teeth (100–77% = 23%) and we should be very cautious for

these endodontically treated teeth with negative results.

The limitations of this meta-analysis were small sample

sizes in some studies, no applying reference standard test for

all patients in some studies, and unavailability of data for

subgroup analysis for horizontal and oblique tooth fractures.

Moreover, as mentioned in Table 2, CBCT devices and

exposure protocols differed among included studies. Since

image quality may vary among different CBCT devices and

exposure protocols,44,45 the results in this meta-analysis

should be interpreted with caution and may not be applied

to all CBCT devices.



Fig. 5 – A recommended diagnostic pathway for patients

with clinically suspected tooth fractures.



5. Clinical implications

According to European Guidelines (Radiation Protection 172:

cone beam CT for dental and maxillofacial radiology)46 and the

results in this meta-analysis, we suggest the following diagnos-

tic pathway for patients with clinically-suspected tooth frac-

tures (Fig. 5). A patient with clinically-suspected tooth fractures

should be firstly diagnosed through periapical radiography. If

tooth fractures are detected by periapical radiography, subse-

quent treatments should follow; if tooth fractures are not

detected by the periapical radiography but symptoms persist,

CBCT should be prescribed for further diagnosis. If tooth

fractures are found by CBCT, regardless of endodontically or

non-endodontically treated teeth, treatments should ensue; if

tooth fractures are not found by CBCT, we should be very

cautious due to chances of false negatives and thus close follow-

ups are needed. Particularly, special attention should be given to

endodontically treated teeth with negative CBCT findings.

Furthermore, in this meta-analysis, almost all the included

participants were adults, thus the results should be limited to

adults. Although CBCT exams for tooth fractures enjoy the

aforementioned advantages, due to radiation risks, we suggest

that CBCT exams for tooth fractures should be used unless

good justifications are made (tooth fractures cannot be

detected through periapical radiography). Moreover, radiation

risks of CBCT are higher in children than in adults,46 thus we

should be very cautious for prescribing CBCT exams in

children. For the diagnosis of tooth fractures through CBCT,

small and medium field of view (FOV) (effective dose: 11–

674 mSV) should be used rather than large FOV (effective dose:

30–1073 mSV) for radiation protection.46

6. Conclusion

We suggest that CBCT has a high diagnostic accuracy for tooth

fractures and could be used in clinical settings. We can be very

confident with positive test results but should be very cautious

with negative test results. For patients with negative results,

close follow-ups are recommended.

The diagnostic accuracy of CBCT is similar among different

types of tooth fractures, which should be interpreted with


caution due to unavailability of data for subgroup analysis on

horizontal and oblique tooth fractures.

The diagnostic accuracy of CBCT is higher in non-

endodontically treated teeth than in endodontically treated

teeth. We can be very confident with the positive results for

both but should be cautious with the negative results

especially for endodontically treated teeth.

Clinical significance

Through an extensive meta-analysis, this systematic review

critically examines the diagnostic performance of CBCT for

tooth fractures and provides an evidence-based diagnostic

pathway for tooth fractures. This diagnostic pathway would

be beneficial for practitioners to make correct diagnosis and

prudent treatment decisions.

Acknowledgement

This work was supported by National Natural Science

Foundation of China (NSFC), Nos. 81070858 and 81100778.

r e f e r e n c e s

1. Chai H, Lee JJ, Constantino PJ, Lucas PW, Lawn BR.Remarkable resilience of teeth. Proceedings of the NationalAcademy of Sciences of the United States of America2009;106:7289–93.

2. Gher Jr ME, Dunlap RM, Anderson MH, Kuhl LV. Clinicalsurvey of fractured teeth. Journal of the American DentalAssociation 1987;114:174–7.

3. Fennis WM, Kuijs RH, Kreulen CM, Roeters FJ, Creugers NH,Burgersdijk RC. A survey of cusp fractures in a population ofgeneral dental practices. International Journal of Prosthodontics2002;15:559–63.

4. Burke FJ. Tooth fracture in vivo and in vitro. Journal ofDentistry 1992;20:131–9.

5. Norton E, O’Connell AC. Traumatic dental injuries and theirassociation with malocclusion in the primary dentition ofIrish children. Dental Traumatology 2012;28:81–6.

6. Kocadereli I, Tasman F, Guner SB. Combined endodontic–orthodontic and prosthodontic treatment of fractured teeth.Case report. Australian Dental Journal 1998;43:28–31.

7. Moule AJ, Kahler B. Diagnosis and management of teethwith vertical root fractures. Australian Dental Journal1999;44:75–87.

8. Edlund M, Nair MK, Nair UP. Detection of vertical rootfractures by using cone-beam computed tomography: aclinical study. Journal of Endodontics 2011;37:768–72.

9. Kajan ZD, Taromsari M. Value of cone beam CT in detection ofdental root fractures. Dentomaxillofacial Radiology 2012;41:3–10.

10. Metska ME, Aartman IH, Wesselink PR, Ozok AR. Detectionof vertical root fractures in vivo in endodontically treatedteeth by cone-beam computed tomography scans. Journal ofEndodontics 2012;38:1344–7.

11. Pauwels R, Beinsberger J, Collaert B, Theodorakou C, RogersJ, Walker A, et al. Effective dose range for dental cone beamcomputed tomography scanners. European Journal ofRadiology 2012;81:267–71.

12. Avsever H, Gunduz K, Orhan K, Uzun I, Ozmen B, Egrioglu E,et al. Comparison of intraoral radiography and cone-beam


http://refhub.elsevier.com/S0300-5712(13)00324-2/sbref0005








































computed tomography for the detection of horizontal rootfractures: an in vitro study. Clinical Oral Investigations 2013.

13. da Silveira PF, Vizzotto MB, Liedke GS, da Silveira HL,Montagner F, da Silveira HE. Detection of vertical rootfractures by conventional radiographic examination andcone beam computed tomography – an in vitro analysis.Dental Traumatology 2013;29:41–6.

14. Ferreira RI, Bahrami G, Isidor F, Wenzel A, Haiter-Neto F,Groppo FC. Detection of vertical root fractures by cone-beamcomputerized tomography in endodontically treated teethwith fiber-resin and titanium posts: an in vitro study. OralSurgery Oral Medicine Oral Pathology and Oral Radiology2013;115:e49–57.

15. Gunduz K, Avsever H, Orhan K, Celenk P, Ozmen B,Cicek E, et al. Comparison of intraoral radiography andcone-beam computed tomography for the detection ofvertical root fractures: an in vitro study. Oral Radiology2013;29:6–12.

16. May JJ, Cohenca N, Peters OA. Contemporary managementof horizontal root fractures to the permanent dentition:diagnosis-radiologic assessment to include cone-beamcomputed tomography. Pediatric Dentistry 2013;35:120–4.

17. Dolekoglu S, Fisekcioglu E, Ilguy M, Ilguy D. The usage ofdigital radiography and cone beam computed tomographyamong Turkish dentists. Dentomaxillofacial Radiology2011;40:379–84.

18. Kamburoglu K, Kursun S, Akarslan ZZ. Dental students’knowledge and attitudes towards cone beam computedtomography in Turkey. Dentomaxillofacial Radiology2011;40:439–43.

19. Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ,Reitsma JB, et al. QUADAS-2: a revised tool for the qualityassessment of diagnostic accuracy studies. Annals of InternalMedicine 2011;155:529–36.

20. Mills EJ, Nachega JB, Buchan I, Orbinski J, Attaran A, Singh S,et al. Adherence to antiretroviral therapy in sub-SaharanAfrica and North America: a meta-analysis. Journal of theAmerican Medical Association 2006;296:679–90.

21. Miller JJ. The Inverse of the Freeman–Tukey double arcsinetransformation. American Statistician 1978;32:138.

22. Ingelfinger JA, Mosteller F, Thibodeau LA, Ware JH.Biostatistics in clinical medicine. New York: McGraw-Hill;1994: 26–50.

23. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Fixed-effect versus random-effects models. Introduction to meta-analysis. 1st ed. Cornwall: Wiley; 2009: 77–85.

24. Deeks JJ, Macaskill P, Irwig L. The performance of tests ofpublication bias and other sample size effects in systematicreviews of diagnostic test accuracy was assessed. Journal ofClinical Epidemiology 2005;58:882–93.

25. Zamora J, Abraira V, Muriel A, Khan K, Coomarasamy A.Meta-DiSc: a software for meta-analysis of test accuracydata. BMC Medical Research Methodology 2006;6:31.

26. Du Y, Tang K, Yu X, Lu J. Cone-beam CT application in thediagnosis of tooth longitudinal fractures. Chinese Journal ofConservative Dentistry 2010;20:450–3.

27. Liu R, Gao F, Li Y, Shi X, Li Y. Application of cone beamcomputed tomography (CBCT) in diagnosis and treatment ofdental root fracture. Medical Journal of Chinese People’sLiberation Army 2010;35:1484–5.

28. Ning F, Yu Q, Cao D. Application of cone-beam computedtomography in the early diagnosis of vertical root fracture.Chinese Computerized Medical Imaging 2011;17:506–8.

29. Qin H, Zhang C, Yan B, Fan W, Zhao Q. The abilityof cone-beam CT in the diagnosis of vertical toothfractures. Journal of Zhengzhou University (Medical Sciences)2011;46:933–5.


30. Wang P, Yan XB, Lui DG, Zhang WL, Zhang Y, Ma XC.Detection of dental root fractures by using cone-beamcomputed tomography. Dentomaxillofacial Radiology2011;40:290–8.

31. Xue Y, Ni L, Wang H, Guo C, You S. A comparison of cone-beam computed tomography with periapical radiography indetection of vertical root fracture. Chinese Journal ofConcervative Dentistry 2011;21:233–6.

32. Yan X, Peng H, Liu L, Song A, Sun F. Application of cone-beamcomputed tomography in the early diagnosis of vertical rootfracture. Journal of Practical Stomatology 2011;27:46–9.

33. Ding S. Diagnostic value of cone-beam CT for root fractures.Journal of Mudanjiang Medical University 2012;33:52–4.

34. Liu J, Zhang H, Wu J, Li S, Huang L, Yang P. Comparison ofdiagnostic value of cone beam computed tomography andorthpantomography for root fracture. International Journal ofStomatology 2013;40:17–9.

35. Wacker C, Prkno A, Brunkhorst FM, Schlattmann P.Procalcitonin as a diagnostic marker for sepsis: a systematicreview and meta-analysis. Lancet Infectious Diseases2013;13:426–35.

36. Irwig L, Tosteson AN, Gatsonis C, Lau J, Colditz G, ChalmersTC, et al. Guidelines for meta-analyses evaluating diagnostictests. Annals of Internal Medicine 1994;120:667–76.

37. Irwig L, Macaskill P, Glasziou P, Fahey M. Meta-analyticmethods for diagnostic test accuracy. Journal of ClinicalEpidemiology 1995;48:119–30. discussion 31–2.

38. Walter SD. Properties of the summary receiver operatingcharacteristic (SROC) curve for diagnostic test data. Statisticsin Medicine 2002;21:1237–56.

39. Schmidt RL, Hunt JP, Hall BJ, Wilson AR, Layfield LJ. Asystematic review and meta-analysis of the diagnosticaccuracy of frozen section for parotid gland lesions.American Journal of Clinical Pathology 2011;136:729–38.

40. Jaeschke R, Guyatt GH, Sackett DL. Users’ guides to themedical literature. III. How to use an article about adiagnostic test. B. What are the results and will they helpme in caring for my patients? The Evidence-Based MedicineWorking Group. Journal of the American Medical Association1994;271:703–7.

41. Egger M, Smith GD, Altman DG. Systematic reviews ofevaluations of diagnostic and screening tests. Systematicreviews in health care: meta-analysis in context. London: BMJBooks; 2001: 248–82.

42. Hassan B, Metska ME, Ozok AR, van der Stelt P, WesselinkPR. Detection of vertical root fractures in endodonticallytreated teeth by a cone beam computed tomography scan.Journal of Endodontics 2009;35:719–22.

43. Patel S, Brady E, Wilson R, Brown J, Mannocci F. Thedetection of vertical root fractures in root filled teeth withperiapical radiographs and CBCT scans. InternationalEndodontic Journal 2013.

44. Liang X, Jacobs R, Hassan B, Li L, Pauwels R, Corpas L, et al. Acomparative evaluation of Cone beam computedtomography (CBCT) and multi-slice CT (MSCT) part I. Onsubjective image quality. European Journal of Radiology2010;75:265–9.

45. Pauwels R, Beinsberger J, Stamatakis H, Tsiklakis K, WalkerA, Bosmans H, et al. Comparison of spatial and contrastresolution for cone-beam computed tomography scanners.Oral Surgery Oral Medicine Oral Pathology and Oral Radiology2012;114:127–35.

46. European Commission. Radiation Protection N8 172: conebeam CT for dental and maxillofacial radiology.. 2012.Available from: http://ec.europa.eu/energy/nuclear/radiation_protection/doc/publication/172.pdf[cited 26.10.2013].






































































































































http://ec.europa.eu/energy/nuclear/radiation_protection/doc/publication/172.pdf

http://ec.europa.eu/energy/nuclear/radiation_protection/doc/publication/172.pdf


Documents

Diagnostic accuracy of CBCT for tooth fractures: A meta-analysis