CLINICAL RESEARCH

Recommended treatment of cracked teeth: Results from the

National Dental Practice-Based Research Network
Thomas J. Hilton, DMD, MS,a Ellen Funkhouser, DrPH,b Jack L. Ferracane, PhD,c
Michele Schultz-Robins, BA, DMD,d Valeria V. Gordan, DDS, MS, MS-CI,e Bobby J. Bramblett, DMD,f
R. Mack Snead, Jr, DDS, MAGD,g Walter Manning, DMD,h Jeffrey R. Remakel, DDS,i and
National Dental PBRN Collaborative Group

ABSTRACT
Statement of problem. Despite the high prevalence of posterior cracked teeth, questions remain regarding the best course of action for
managing these teeth.
Purpose. The purpose of this clinical study was to identify and quantify the characteristics of visible cracks in posterior teeth and their
association with treatment recommendations among patients in the National Dental Practice-Based Research Network.
Material and methods. Network dentists enrolled patients with a single, vital posterior tooth with at least 1 observable external crack. Data
were collected at the patient, tooth, and crack levels, including the presence and type of pain and treatment recommendations for subject
teeth. Frequencies according to treatment recommendation were obtained, and odds ratios (ORs) comparing recommendations for the tooth
to be restored versus monitored were calculated. Stepwise regressions were performed using generalized models to adjust for clustering;
characteristics with P<.05 were retained.
Results. A total of 209 dentists enrolled 2858 patients with a posterior tooth with at least 1 crack. Mean ±standard deviation patient age was
54 ±12 years; 1813 (63%) were female, 2394 (85%) were non-Hispanic white, 2213 (77%) had some dental insurance, and 2432 (86%) had
some college education. Overall, 1297 (46%) teeth caused 1 or more of the following types of pain: 1055 sensitivity to cold, 459 biting,
and 367 spontaneous. A total of 1040 teeth were recommended for 1 or more treatments: restoration (n=1018; 98%), endodontics (n=29;
3%), endodontic treatment and restoration (n=20; 2%), extraction (n=2; 0.2%), and noninvasive treatment, for example, occlusal device,
desensitizing (n=11; 1%). The presence of caries (OR=67.3), biting pain (OR=7.3), and evidence of a crack on radiographs (OR=5.0) were
associated with over 5-fold odds of recommending restoration. Spontaneous pain was associated with nearly 3-fold odds; pain to cold,
having dental insurance, a crack that was detectable with an explorer or blocked transilluminated light, or connected with a restoration
were each weakly associated with increased odds of recommending a restoration (OR<2.0).
Conclusions. Approximately one-third of cracked teeth were recommended for restoration. The presence of caries, biting pain, and
evidence of a crack on a radiograph were strong predictors of recommending a restoration, although the evidence of a crack on a
radiograph only accounted for a 3% absolute difference (4% recommended treatment versus 1% recommended monitoring). (J Prosthet
Dent 2019;-:---)

Supported by NIDCR (grant U19-DE-22516).

The National Dental PBRN Collaborative Group comprises practitioner, faculty, and staff investigators who contributed to this network activity. A list of these persons is at
http://www.nationaldentalpbrn.org/collaborative-group.php.
a
Alumni Centennial Professor in Operative Dentistry, School of Dentistry, Oregon Health & Science University, Portland, Ore.
b
Associate Professor, School of Medicine, University of Alabama, Birmingham Ala.
c
Chair, Department of Restorative Dentistry, School of Dentistry, Oregon Health & Science University, Portland, Ore.
d
Clinical Assistant Professor, Restorative Department, Rutgers School of Dental Medicine, Newark, NJ.
e
Professor, Department of Restorative Dental Sciences, University of Florida, Gainesville, Fla.
f
Private practice, Leeds, Ala.
g
Private practice, Fort Worth, Texas.
h
Private practice, Albany, Ore.
i
Private practice, St Louis Park, Minn.

THE JOURNAL OF PROSTHETIC DENTISTRY 1

2 Volume - Issue -

previous publication.22 In brief, the study used a conve-

Clinical Implications nience sample of participants between the age of 19 and
Various posterior tooth characteristics affect the 85 years enrolled by dentists in the National Dental
Practice-based Research Network.23 To be eligible, par-
clinician’s decision to monitor versus treat a cracked
ticipants were required to have at least 1 single, vital
tooth. When the decision is made to treat the tooth,
posterior tooth with at least 1 observable external crack.
most commonly because of the practitioner’s
Participating dentists were asked to select 1 of these teeth
concern about the integrity of the tooth or because
in each patient and to characterize this tooth for 20
the patient experienced pain, dentists and patients
eligible participants or as many as they could enroll in 8
will usually opt for restorative treatment. The
weeks, whichever came first. The study was reviewed and
restorative treatment of choice for posterior cracked
teeth is a complete crown. approved by the institutional review board of the lead
investigators (T.J.H., J.L.F.), as well as by the various
institutional review boards that oversee the 6 regions of
the network. Informed consent was obtained for all
Teeth with cracks are a common occurrence in adults,
participants enrolled into the study. There were 2 phases
with prevalence rates of up to 70%, depending on the
to enrollment: a pilot phase with 183 patients from 12
tooth type and location.1 The diagnosis and treatment of
practices from April through July 2014, and a main
cracked teeth have been challenging for dentists and
launch phase that occurred from October 2014 through
patients, and the outcomes can be consequential, with
April 2015. The study instruments were evaluated in the
the need for a major restoration, root canal therapy, or
pilot phase to address understandability, coverage, and
extraction.2 As a result, finding the best treatment option
ease of form completion and were revised for full-study
for cracked teeth is a priority. Various procedures have
implementation based on feedback from pilot
been suggested either to aid in the diagnosis or treatment
practitioners.
of a cracked tooth, including occlusal adjustment, seda-
Dentists and their designated practice personnel
tive interim restorations, placement of orthodontic bands,
were trained in data collection using a training manual
interim crowns, direct or indirect composite resin resto-
developed and approved by the study principal in-
rations, complex and bonded amalgam restorations, and
vestigators (PIs). Data including the presence and type
partial and complete indirect crowns.3-12
of pain, as well as data on treatment recommendations
In a practice-based study in which 1777 dentists were
for participant teeth, were collected at the patient,
presented with various clinical scenarios, the presence of
tooth, and crack levels. Data forms are publicly available
a crack or fracture was the factor most likely to result in
at http://nationaldentalpbrn.org/study-results/cracked-
the dentist recommending a crown.11 Another study
tooth-registry.php. Confirmation of tooth vitality of
presented 95 dentists (generalists, prosthodontists, and
enrolled teeth was with cold24 (for example, refrigerant
endodontists) with 4 different clinical cracked tooth
or ice), although other methods were used such as air,
scenarios and asked what treatment they would recom-
air-water spray, or electric pulp testing. Spontaneous
mend. Treatment suggestions were wide-ranging and
pain information was obtained from patient report;
were not related to the practitioners’ specialty. A variety
sensitivity to cold was ascertained using refrigerant, ice,
of factors contribute to the decision to restore versus
or air-water spray; and pain upon biting was verified by
monitor a tooth, including caries, the quality of the
having the patient occlude on a device or instrument
remaining tooth structure, presence of a visible fracture
placed on the occlusal surface of the cracked tooth. To
line, sealing the tooth against bacterial ingress, protecting
help patients discriminate between pain (an increased
cusps against flexure under function, presence of an
response to the cold or bite assessment) and an ordi-
existing restoration, and whether or not the patient has
nary response, dentists were asked to also perform
dental insurance.12-21 Evidence-based guidelines are
these tests on an unaffected (for example, the contra-
needed for the treatment of cracked teeth. The purpose
lateral) tooth. Practitioners indicated reason(s) why they
of this clinical study was to contribute to this evidential
recommended the study teeth for treatment from a list
foundation by identifying and quantifying characteristics
of 9 options (with the instruction to check all that apply,
at the patient, tooth, and crack levels and their associa-
plus the option to write in an additional reason). If a
tion with treating posterior teeth with visible cracks
practitioner recommended a tooth for restoration, they
among patients enrolled in the National Dental Practice-
were asked to specify restoration type (intracoronal,
Based Research Network.
partial crown, or complete crown), placement technique
(direct or indirect), and adhesive bonding (yes or no).
MATERIAL AND METHODS
Frequencies according to treatment recommendation
A detailed report of the study procedures, including were obtained, and odds ratios (ORs) were calculated for
enrollment and data collection, has been provided in a recommendations of treatment versus monitoring. As the

THE JOURNAL OF PROSTHETIC DENTISTRY Hilton et al

- 2019 3

Radiograph 4%
1%
Factors in Restore versus Monitor Decision

Connects with restoration 78%

71%

Dental insurance 83%

74%

Blocks transilluminated light 71%

62%

Detectable with explorer 76%

66%

25%
Spontaneous pain 6%

Cold pain 50%

29%

Biting pain 34%

6%

Caries present 0% 29%

0% 20% 40% 60% 80% 100%

Proportion of Each Restore vs. Monitor Factor*
Restore (N=1018) Monitor (N=1818)

Percentages are proportions of each factor according to whether decision was to restore or monitor.

Figure 1. Distribution of factors in dentists’ decision to restore or monitor cracked teeth (N=2836).

validity of a statistical test depends on independent ob- for clustering of patients within the practice were entered
servations and the model and as the test must reflect the into a full model, followed by backward elimination to
correlation structure of the data to yield valid estimates of remove all variables for which P was .05, using GEE to
variance and valid statistical tests, patients within a adjust for clustering. After fitting the final model, all
practice represent clusters that are often correlated to the interaction terms were tested for significance at the 5%
outcome being studied.25 Clustering typically reduces level. To assess the robustness of findings, regressions
precision of estimation, yielding lower statistical power were repeated, comparing all definitive treatment rec-
and wider confidence intervals than studies of equal ommendations (extraction, endodontics, and restora-
sample size but without clusters. In a univariable fashion, tions) to monitoring only. All ORs and P values were
each patient-, tooth-, and crack-level characteristic was adjusted for clustering of patients treated by the same
entered into a logistic regression model that used practitioner with GEE. All analyses were performed using
generalized estimating equations (GEEs) adjusted for statistical software (SAS v9.4; SAS Institute Inc).
clustering of patients within the practice and imple-
mented using the SAS procedure for generalized models,
RESULTS
with an exchangeable compound symmetric correlation
matrix (PROC GENMOD in SAS with the CORR=EXCH A total of 2858 patients with a posterior cracked tooth
option). This approach specified a model in which ob- were enrolled by 209 practitioners. The mean/median
servations on individual patients seen by a particular was 14.8/15 patients per practice, and the range was 1 to
practitioner are allowed to be correlated, whereas those 20. The distribution of the characteristics that study
from different practitioners are assumed to be indepen- dentists took into consideration when deciding whether
dent. This approach removed variability caused by dif- to restore versus monitor a cracked tooth is presented in
ferences among practitioners from the tests for Figure 1.
association between the predictor variable and the A total of 1040 teeth (36%) were recommended for
outcome variable and so uses the appropriate estimate of the following treatments: restoration only (998; 96%),
standard error for statistical tests. endodontic treatment only (9; 0.1%), endodontic treat-
To identify independent associations for recom- ment and restoration (20; 2%), extraction (2; 0.2%), or
mending that a study tooth be restored versus moni- noninvasive treatment (for example, occlusal device,
tored, all characteristics with P<.05 after adjusting only desensitizing [11; 1%]). The disposition of the 1018

Hilton et al THE JOURNAL OF PROSTHETIC DENTISTRY

4 Volume - Issue -

100 Table 1. Analysis of tooth-level characteristics according to treatment

versus monitor recommendation at baseline
80 Not bonded, 38 Monitor Restore
Indirect, 55 (N=1818) (N=1018)
Complete crown, 61
Cracks (%)

60 Tooth-Level Characteristic Na Col%b N Row%c

Molar 1436 79 877 38
40 Premolar 382 21 141 27
Partial crown, 3
Bonded, 62 Cluster-adjusted ORd OR=1.6
20 Intracoronal, 35
Direct, 45 Cluster-adjusted Pe P<.001
2 or more external cracks 680 37 691 50
0 1 external crack 1138 63 327 22
Restoration type Placement type Bonded
Cluster-adjusted OR OR=1.3
Treatment Type Cluster-adjusted P P=.006
Missing: restoration type, n=14; placement type, n=4; whether or not bonded, n=4
Wear facet through enamel 385 21 286 43
Figure 2. Restorative treatment recommendations by restoration type, No wear facet through enamel 1433 79 732 34
placement type, and bonding (N=1018). Cluster-adjusted OR OR=1.4
Cluster-adjusted P P<.001
Exposed roots 427 23 206 33
cracked teeth recommended for restoration is as
No exposed roots 1391 77 812 37
followsdtype of restoration: 357 (35%) intracoronal, 34
Cluster-adjusted OR OR=0.8
(3%) partial crown, and 623 (61%) complete crown; type Cluster-adjusted P P=.018
of placement: direct 452 (45%) and indirect 562 (55%); Caries present 6 <1 295 98
bonded: yes 624 (62%) and no 380 (38%) (Fig. 2). No caries present 1812 100 723 29
Virtually all those teeth recommended for indirect Cluster-adjusted OR OR=54.8
placement were to receive complete crowns (N=534; Cluster-adjusted P P<.001
95%), and the majority of those teeth recommended for NCCL present 169 9 83 33
direct placement were to receive an intracoronal resto- No NCCL present 1649 91 935 36
ration (N=355; 79%). Similarly, the majority of restora- Cluster-adjusted OR OR=.8
tions not recommended for bonding were complete Cluster-adjusted P P=.122
crowns (N=315; 83%). Approximately equal numbers of Evidence of crack(s) on radiograph 12 1 41 77
intracoronal restorations and complete crowns were No evidence of crack(s) on radiograph 1806 99 977 35
Cluster-adjusted OR OR=4.9
recommended as bonded restorations.
Cluster-adjusted P P<.001
a
Practitioner-level characteristics NCCL, noncarious cervical lesion; OR, odds ratio. Column Ns not summing to column
total N above due to missing data. bColumn percentages not summing to 100 due to
The mean age ±standard deviation of the practitioners rounding. cPercentage recommended for restoration within level of tooth-level
was 53 ±10; the median age (interquartile range) was 56 characteristic: (# with recommend restore)/(# recommend restore + # recommend
monitor). dOR: Odds ratio adjusted only for clustering of patients within a practitioner
(45 to 60) years, with a range from 27 to 73 years. Of the using generalized estimating equations. Typically, cluster-adjusted OR is similar to crude
209 dentists participating in the study, 153 (73%) prac- OR, for example, molar crude OR=(877)(382)/(1436)(141)=1.6, same as cluster-
adjusted OR. In contrast, for multiple cracks, crude OR=(691)(1138)/(680)(327)=3.5,
titioners were male and 173 (83%) were non-Hispanic whereas adjusted OR is lower at 1.3. For caries, crude OR=(295)(1812)/(6)(723)=123
white. Two practitioners were periodontists, and the with adjusted OR=54.8. Difference between crude OR and adjusted OR happens when
patients of few practitioners differ from patients of majority. Large difference for caries
other 207 were general practitioners. Over half (N=118; partly due to small number (6) of patients with caries monitored and not restored.
e
56%) were solo, private practitioners, with almost Significance of differences in proportions recommended to restore adjusted only for
clustering of patients within practitioner using generalized estimating equations.
another third being either owners of nonesolo private
practice settings (N=46; 22%) or associates in private
practices (N=17, 8%). Thirteen (6%) were in large group 66%) reported clenching, grinding, or pressing their teeth
practices offering preferred care (HealthPartners or Per- together, and 2190 of 2690 main launch participants
manente Dental Associates), and 6 (3%) were in aca- (81%) reported feeling at least some stress, with over
demic centers. one-third reporting feeling stressed at least weekly
(N=1048, 39%). Data on stress were not obtained in the
Patient-level characteristics pilot phase. A total of 1297 (45%) teeth were symp-
The age range of patients was 19 to 85 years, with a mean tomatic. Pain was noted from cold stimuli (N=1055; 81%)
age ±standard deviation of 54 ±12 years and a median and biting (N=459; 35%); spontaneous pain was also
age (interquartile range) of 55 (46 to 62) years. Of the reported (N=367; 28%), and 409 (35%) had more than 1
2858 patients enrolled in the study, 1813 (63%) were type of symptom.
female, 2394 were non-Hispanic white (83%), 2213 had The age was inversely associated with a tooth being
some dental insurance (77%), and 2432 had some college recommended for restoration (OR=0.86 per 10 years,
education (85%). Two-thirds of the patients (N=1900, P<.001). A patient who had dental insurance (OR=1.4,

THE JOURNAL OF PROSTHETIC DENTISTRY Hilton et al

- 2019 5

Table 2. Analysis of crack-level characteristics according to treatment Table 3. Independent associations with cracked tooth being
versus monitor recommendation at baseline recommended for restoration versus monitoringa
Monitor Restore 95% Confidence
(N=1818) (N=1018) Characteristic Odds Ratio Interval Pb
Crack-Level Characteristics Na Col%b N Row%c Caries present 67.9 37.6-122.6 <.001
At least 1 crack stained 1464 81 843 37 Biting pain 7.3 5.2-10.2 <.001
No cracks were stained 354 19 175 33 Evidence on radiograph 4.8 2.6-8.8 <.001
Cluster-adjusted ORd OR=1.3 Spontaneous pain 2.9 2.0-4.0 <.001
Cluster-adjusted Pe P=.006 Cold pain 1.7 1.4-2.2 <.001
At least 1 crack detectable with an explorer 1192 66 773 39 Dental insurance 1.3 1.1-1.6 .006
No cracks detectable with an explorer 626 34 245 28 Has a crack that
Cluster-adjusted OR OR=1.8 Is detectable with explorer 1.6 1.2-2.0 <.001
Cluster-adjusted P P<.001 Blocks transilluminated light 1.4 1.1-1.8 .019
At least 1 crack blocked transilluminated light 1126 62 726 39 Connects with restoration 1.4 1.1-1.8 .005
No cracks blocked transilluminated light 692 38 292 30 Is in horizontal direction 1.3 1.0-1.6 .024
Cluster-adjusted OR OR=1.6 a
Teeth recommended for nonsurgical treatments (n=11), endodontics only (n=9), or extraction
Cluster-adjusted P P<.001 (n=2) are excluded. bFrom generalized estimating equations adjusting for clustering
within practice using stepwise regression, retaining if P<.05.
At least 1 crack connected with a restoration 1285 71 794 38
No cracks connected with a restoration 533 29 224 30
The presence of caries was strongly associated with a
Cluster-adjusted OR OR=1.4
Cluster-adjusted P P<.001
tooth being recommended for restoration rather than
At least 1 crack connected with another crack 106 6 97 48
monitoring (OR=54.8, P<.001). Evidence of a crack on a
No cracks connected with another crack 1712 94 921 35 radiograph was also strongly associated with a restora-
Cluster-adjusted OR OR=1.5 tion recommendation (OR=4.9, P<.001), whereas a crack
Cluster-adjusted P P=.023 on a molar (OR=1.6, P<.001), multiple external cracks
At least 1 crack extended to root 308 17 219 42 (OR=1.3, P=.006), and the presence of a wear facet
No cracks extended to root 1510 83 799 35 through the enamel (OR=1.4, P<.001) were each
Cluster-adjusted OR OR=1.0 modestly associated with a recommendation for resto-
Cluster-adjusted P P=.771 ration. Cracked teeth with exposed roots were inversely
a
OR, odds ratio. Column Ns not summing to column total N above due to missing data. associated (OR=0.8, P=.018) with a restoration
b
Column percentages not summing to 100 due to rounding. cPercentage recommended
for restoration within level of crack-level characteristic: (# with recommend restore)/
recommendation (Table 1).
(# recommend restore + # recommend monitor). dOR: Odds ratio adjusted only for
clustering of patients within practitioner using generalized estimating equations.
e
Significance of differences in proportions recommended to restore adjusted only for
Crack-level characteristics
clustering of patients within practitioner using generalized estimating equations. Overall, study teeth exhibited the following crack-level
characteristics: stained (N=2319; 81%), connected with a
restoration (N=2095; 73%), detectable with an explorer
P<.001), cold pain (OR=2.8, P<.001), biting pain (N=1980; 69%), or blocked transilluminated light (N=1862;
(OR=9.0, P<.001), and spontaneous pain (OR=5.6, 65%). Fewer teeth presented with the crack extending to the
P<.001) were likely to be recommended to receive a root (N=297; 10%) or connected with another crack (N=121;
restoration rather than monitoring. 4%). Tooth surfaces with cracks varied over a narrow range,
from 44% (N=1267) involving the occlusal surface to 51%
Tooth-level characteristics (N=1463) involving the lingual surface; 1028 (36%) had a
Most cracked teeth were molars (N=2332; 82%), and crack that involved 2 or more surfaces.
more than half of these were in the mandibular arch Cracks that stained (OR=1.3, P=.006), that were
(N=1675, 59%). The vast majority of external cracks, detectable with an explorer (OR=1.8, P<.001), that
N=2640 (92%), were on a tooth with a restoration: blocked transilluminated light (OR=1.6, P<.001), that
N=2041 (71%) of cracked teeth had 1 restoration, N=547 connected with a restoration (OR=1.4, P<.001), or
(19%) had 2 restorations, and N=52 (2%) had 3 or 4 that connected with another crack (OR=1.5, P=.023) were
restorations. Slightly more than one-third (N=1018; each modestly associated with an increased likelihood of
36%) of teeth had 1 external crack, 759 (27%) had 2, 507 the tooth being recommended for restoration (Table 2).
(18%) had 3, and 574 (20%) had 4 or more. Of the total,
638 (22%) had some root exposure, 676 (24%) presented Independent associations
with at least 1 wear facet through enamel, and 254 (9%) All possible 2-way interactions were evaluated, and none
had a noncarious cervical lesion. Only 53 (2%) had evi- were found significant, indicating that an additive model
dence of a crack on a radiograph. Of 302 (11%) teeth is sufficient. The independent associations with a tooth
with caries, only 6 (<1%) were on a tooth that practi- being recommended for restoration versus monitoring in
tioners recommended for monitoring. the final models are presented in Table 3.

Hilton et al THE JOURNAL OF PROSTHETIC DENTISTRY

6 Volume - Issue -

Each Restoration Recommendation (%)

0 20 40 60 80 100

Compromised tooth structure 64

Reasons for Recommending Restoration

Sensitive and/or painful 35

Broken/defective restoration 31

Sensitive to hot or cold 25

Caries associated with crack 24

Painful or infected 18

Caries NOT associated with crack 14

Figure 3. Reasons for recommendation of restorative treatment (N=1018).

The presence of caries (OR=67.3; P<.001), biting pain DISCUSSION

(OR=7.3; P<.001), and evidence of a crack on radiographs
This study showed that if active intervention for a pos-
(OR=4.8; P<.001) were each strongly associated with
terior cracked tooth (versus monitoring) is chosen, almost
recommending restoration. Spontaneous pain was
always (98%), the management of choice is restorative
associated with a nearly 3-fold increase in the odds of the
treatment. Several factors go into the decision to restore
tooth being recommended for treatment (OR=2.9; P<.001).
versus monitor a cracked tooth. In this study, factors that
Pain to cold (OR=1.7; P<.001); having a crack that was
were most important in deciding to restore a cracked
detectable with an explorer (OR=1.6; P<.001), that blocked
tooth were the presence of caries, pain, or radiographic
transilluminated light (OR=1.4; P=.019), that connected
evidence of a crack. Previous studies have reported a
with a restoration (OR=1.4; P=.005), and that extended in a
significant correlation between caries and radiographic
horizontal direction (OR=1.3; P=.024); and a patient with
evidence of a crack and pain in cracked teeth.12,13 A
dental insurance (OR=1.3; P=.006) were each weakly
strong correlation was found between radiographic evi-
associated with an increased odds of a recommendation
dence of a crack and the practitioner’s decision to restore
for restoration (OR<2.0).
the tooth (OR=4.8), although the radiographic evidence
Analysis including all invasive treatment recommen-
of a crack was low and the risk difference was only 3%.
dations, namely, extraction, endodontic treatment, and
However, when present, an evident crack on a radio-
restoration, compared with monitoring did not differ
graph strongly correlated with pain and often resulted in
from the final model presented in Table 3 comparing
a recommendation from the practitioner to restore the
restoration to monitoring, excluding extraction and
tooth. Pain would dictate that a practitioner take defini-
endodontic treatment. Numbers were insufficient to
tive action to relieve discomfort. Likewise, caries is the
analyze extraction (n=2) separately. When modeling
most prevalent disease in both children and adults14 and
endodontic treatment either with or without restoration,
as such mandates that therapy be instituted to address it.
only spontaneous pain (OR=13.0, 95% confidence in-
When caries was detected in conjunction with a cracked
terval [CI]: 4.2 to 40.5, P<.001), biting pain (OR=7.0, 95%
tooth, the practitioners in this study typically and pre-
CI: 2.8 to 17.6, P=.001), and cold pain (OR=5.5, 95% CI:
dictably selected restorative treatment.
1.99-16.0, P=.006) were independently significantly
Other factors were more equivocal in their impact on a
associated with increased odds of recommending resto-
dentist’s decision to restore or monitor a cracked tooth,
ration. The most common reason noted for recom-
with similar number of dentists weighing factors such as
mending restoration was compromised tooth structure
the crack being detectable with an explorer, blocking
(64%), followed by painful or sensitive teeth (35%),
transilluminated light, and connecting with a restoration,
broken or defective restorations (31%), caries associated
as well as whether the patient had dental insurance, as
with a crack (24%), and caries not associated with a crack
justification for both restoring and monitoring a cracked
(14%) (Fig. 3).

THE JOURNAL OF PROSTHETIC DENTISTRY Hilton et al

- 2019
tooth. This confirms previous data that there is a low level
of concurrence among dentists regarding the best way to
manage a cracked tooth. A survey of 959 dentists asking
them to rate the importance of 8 factors in cusp fracture
found that, with the exception of dentinal support, no
other factor was rated as very important by a majority of
respondents, and only 1 factor (wear facets) was rated as
important by more than one-third or those participating.15
However, our study also agreed with the survey regarding
the importance of dentists’ assessment of the quality of
remaining tooth structure. The most common rationale for
recommending restorative treatment was the dentist’s
judgment that tooth structure was compromised. This was
cited almost twice as often as the next most common
reason, a sensitive or painful tooth (64% versus 35%).
Bader et al16 determined in a case control study that the 2
leading risk indicators for cusp fracture were the presence
of a fracture line and an increase in the proportional vol-
ume of the natural tooth crown replaced with a restoration,
both of which could be considered as contributing to
compromising the integrity of the tooth.
When the decision was made to restore a cracked tooth,
most of the time (61%), the restoration of choice was a
complete crown. A previous National Dental practice-
based research network (PBRN) study found that the most
common reason for treatment planning a crown was a
tooth with a crack or fracture.11 Many treatments have
been suggested for cracked teeth, ranging from short-term
treatment directed at pain relief and aiding in diagnosis
such as occlusal adjustment, sedative restoration, place-
ment of an orthodontic band or interim crown,6-9 or a
direct composite onlay splint3 to definitive restorations
including direct resin composite,6 indirect resin compos-
ites,4 and crowns.5 The clear preference found in this study
was a crown. The primary rationales for restorative treat-
ment for a cracked tooth are biological, that is, sealing an
avenue of bacterial contamination and toxic element
ingress,4,7,8,17 and mechanical, that is, splinting the frac-
tured elements of the tooth to prevent tooth flexure
causing pain and allowing crack progression.7,18 A com-
plete crown accomplishes both of these functions.
Although associated with lesser ORs, several factors
were found to be associated with a recommendation for
restoring a cracked tooth, including having a crack that
was detectable with an explorer, connected with a
restoration, or blocked transilluminated light and a pa-
tient with dental insurance. A crack that is detectable
with an explorer may be considered to represent a more
definitive break in tooth structure and is potentially more
ominous in terms of adverse events and therefore rec-
ommended for restorative treatment, particularly when
associated with an adjacent restoration.19 A recent
study20 reported that an external crack that connected
with a restoration was more likely to have an internal
crack in that tooth. Although supported by minimal
Hilton et al
7
evidence, the blocking of transilluminated light through a
tooth is often recommended as a method of determining
that a crack penetrates to dentin.2 Cracks that penetrate
to dentin are suggestive of more extensive damage that
would prompt a clinician to plan treatment to protect
against more unfavorable outcomes, such as tooth frac-
ture or a crack impinging on the pulp. A review found
that individuals with dental insurance were more likely to
use dental services than the uninsured; it is therefore
understandable that this would be a factor in recom-
mending treatment for cracked teeth.21
This study has several limitations. For practical reasons,
the study population was not randomly selected. This
allowed participating practitioners to select patients who
both met the inclusion criteria and were most likely to return
for recall visits. Such nonrandom selection could introduce
bias, for example, if study patients are not representative of
individuals who do not enter the dental care system. How-
ever, the long-term goal of the study is to develop guidelines
for those dentists and patients who do participate in regular
dental care. Another potential weakness was the subjective
nature of specific measures used in the study. Although all
participating personnel underwent training before partici-
pating, the fact that some of the assessments do not lend
themselves to purely objective measurement could allow for
variation in recorded data among the participants. Several
clinical measures were used, and these could have been
subject to errors in classification. However, misclassifications
were probably random, and therefore, associations reported
are likely underestimated.
The study strengths include a high number of partici-
pants from a large variety of dental practices across the
United States. These practices collected a large amount of
data in a systematic, controlled manner. Importantly, these
patients will be followed up for several years, hopefully
allowing the assessment of the effectiveness of various
management alternatives for the treatment of cracked teeth.
CONCLUSIONS
Based on the findings of this clinical study, the following
conclusions were drawn:
1. Caries and pain are most likely to result in practi-
tioners recommending treatment of a cracked tooth.
2. Most of the time, the treatment of choice for a
cracked tooth is placement of a restoration, and the
restoration of choice is a complete crown.
REFERENCES
1. Hilton T, Mancl L, Coley Y, Baltuck C, Ferracane J, Peterson J, NW PRE-
CEDENT. Initial treatment recommendations for cracked teeth in Northwest
PRECEDENT. J Dent Res 2011;91(A):Abstract 2387.
2. Lubisich E, Hilton T, Ferracane J. Cracked teeth: a review of the literature.
J Esthet Restor Dent 2010;22:158-67.
THE JOURNAL OF PROSTHETIC DENTISTRY
8 Volume
3. Banerji S, Mehta S, Kamran T, Kalakonda M, Millar B. A multi-centred
clinical audit to describe the efficacy of direct supra-coronal splinting e A
minimally invasive approach to the management of cracked tooth syndrome.
J Dent 2014;42:862-71.
4. Signore A, Benedicenti S, Covani U, Ravera G. A 4- to 6-year retrospective
clinical study of cracked teeth restored with bonded indirect resin composite
inlays. Int J Prosthodont 2007;20:609-16.
5. Krell K, Rivera E. A six-year evaluation of cracked teeth diagnosed with
reversible pulpitis: treatment and prognosis. J Endod 2007;33:1405-7.
6. Opdam N, Roeters J, Loomans B, Bronkhorst M. Seven-year clinical evalu-
ation of painful cracked teeth restored with a direct composite restoration.
J Endod 2008;34:808-11.
7. Fox K, Youngson C. Diagnosis and treatment of the cracked tooth. Prim Dent
Care 1997;4:109-13.
8. Abbot P, Leow N. Predictable management of cracked teeth with reversible
pulpitis. Aust Dent J 2009;54:306-15.
9. Banerji S, Mehta S, Millar B. Cracked tooth syndrome. Part 2: restorative options
for the management of cracked tooth syndrome. Br Dent J 2010;208:503-14.
10. Kim S, Kim S, Cho S, Lee G, Yang S. Different treatment protocols for
different pulpal and periapical diagnoses of 72 cracked teeth. J Endod
2013;39:449-52.
11. McCracken M, Louis D, Litaker M, Minye H, Mungia R, Gordan V, et al.
Treatment recommendations for single-unit crowns. Findings from the national
dental practice-based research network. J Am Dent Assoc 2016;147:882-90.
12. Alkhalifah S, Alkandari H, Sharma P, Moule A. Treatment of cracked teeth.
J Endod 2017;43:1579-86.
13. Hilton TJ, Funkhouser E, Ferracane JL, Gordan VV, Huff KD, Barna J,
et al. Associates of types of pain with crack-level, tooth-level, and
patient-level characteristics in posterior teeth with visible cracks: Find-
ings from the National Dental PBRN Collaborative Group. J Dent
2018;70:67-73.
14. NIDCR. Dental caries. Available at: https://www.nidcr.nih.gov/research/data-
statistics/dental-caries. Accessed May 7, 2018.
15. Bader J, Shugars D, Roberson T. Using crowns to prevent tooth fracture.
Community Dent Oral Epidemiol 1996;24:47-51.
16. Bader J, Shugars D, Martin J. Risk indicators for posterior tooth fracture. J Am
Dent Assoc 2004;135:883-92.
17. Kahler B, Moule A, Stenzel D. Bacterial contamination of cracks in symp-
tomatic vital teeth. Aust Endod J 2000;26:115-8.
18. Seo D, Yi Y, Shin S, Park J. Analysis of factors associated with cracked teeth.
J Endod 2012;38:288-92.
19. Clark D, Sheets C, Paquette J. Definitive diagnosis of early enamel and dentin
cracks based on microscopic evaluation. J Esthet Restor Dent 2003;15:
391-401.
20. Ferracane J, Funkhouser E, Hilton T, Gordan V, Graves C, Giese K, et al.
Observable characteristics coincident with internal cracks in teeth: findings
THE JOURNAL OF PROSTHETIC DENTISTRY
- Issue -
from the National Dental Practice-Based Research Network. J Am Dent
Assoc 2018;149:885-92.e6.
21. Eklund S. The impact of insurance on oral health. J Am Coll Dent 2001;68:
8-11.
22. Hilton TJ, Funkhouser E, Ferracane JL, Gilbert GH, Baltuck C, Benjamin P,
et al. Correlation between symptoms and external cracked tooth character-
istics: findings from the National Dental Practice-Based Research Network.
J Am Dent Assoc 2017;148:246-56.
23. Gilbert GH, Williams OD, Korelitz JJ, Fellows JL, Gordan VV, Makhija SK,
et al. Purpose, structure, and function of the United States National Dental
Practice-Based Research Network. J Dent 2013;41:1051-9.
24. Pigg M, Nixdorf D, Nguyen R, Law A. Validity of preoperative clinical
findings to identify dental pulp status: a National Dental Practice-Based
Research Network study. J Endod 2016;42:935-42.
25. Litaker M, Gordan V, Rindal D, Fellows J, Gilbert G, National Dental PBRN
Collaborative Group. Cluster effects in a National Dental PBRN restorative
study. J Dent Res 2013;92:782-7.
Corresponding author:
Dr Thomas J. Hilton
School of Dentistry
Oregon Health & Science University
2730 S.W. Moody Ave.
Portland, OR 97201-5042
Email: hiltont@ohsu.edu
Acknowledgments
The authors thank the network’s regional coordinators who worked with network
practitioners to conduct the study. Midwest region: Sarah Verville Basile, RDH,
MPH, Christopher Enstad, BS; Western region: Camille Baltuck, RDH, BS, Lisa
Waiwaiole, MS, Natalia Tommasi, MA, LPC; Northeast region: Patricia Ragusa,
BA; South Atlantic region: Deborah McEdward, RDH, BS, CCRP, Brenda Thacker,
AS, RDH, CCRP; South Central region: Claudia Carcelén, MPH, Shermetria
Massengale, MPH, CHES, Ellen Sowell, BA; Southwest region: Stephanie Reyes,
BA, Meredith Buchberg, MPH, Colleen Stewart, MA. The authors are also grateful
to the 12 network practitioners who participated in this study as pilot study
practitioners. Midwest: David Louis, DDS, Timothy Langguth, DDS; Western:
William Reed Lytle, DDS, Don Marshall, DDS; South Atlantic: Stanley Asensio,
DMD, Solomon Brotman, DDS; South Central Region: Jocelyn McClelland, DMD,
James L. Sanderson Jr, DMD; Southwest: Robbie Henwood, DDS, PhD, Michael
Bates, DDS; Northeast: Julie Ann Barna, DMD, MAGD; Sidney Chonowski,
DMD, FAGD.
Copyright © 2019 by the Editorial Council for The Journal of Prosthetic Dentistry.
https://doi.org/10.1016/j.prosdent.2018.12.005
Hilton et al