R E S E A R C H ABSTRACT

Background. Identifying posterior teeth

that are at heightened risk of
developing cusp fracture is A D A
J
an inexact science. Risk

Risk indicators for indicators based on con-

✷

✷

N
CON
trolled observations are

IO
not available, and den-
posterior tooth fracture

T
T

A
N

I
C
tists’ assessments vary. U
A ING EDU 2
Methods. The authors R TICLE
conducted a case-control study
JAMES D. BADER, M.P.H.; DANIEL A. SHUGARS, of cusp fracture in restored posterior teeth.
D.D.S., Ph.D.; JEAN A. MARTIN, D.D.S., M.P.H.
They evaluated 39 potential risk indicators
identified in previous uncontrolled studies
for an association with fracture in 200
e recently reported incidence rates of patients with fractures and 252 patients

W complete cusp fracture in posterior without fractures. These risk indicators

teeth, estimating both overall and tooth delineated patients’ clinical characteristics
type–specific incidence.1 In this report, and behaviors, as well as clinical character-
we continue our examination of pos- istics of individual teeth. The authors used
terior cusp fracture by presenting the results of a case- logistic regression to develop models identi-
control study designed to identify risk indicators for fying risk indicators associated with frac-
such fracture. Information concerning risk indicators is ture, both between case and control subjects
scarce because available studies are all uncontrolled and between case and comparison teeth in
case series, which, by their design, do not permit com- case subjects.
parisons of fractured teeth with sound teeth. Such com- Results. Two risk indicators appeared in
parisons are necessary to establish objectively which both models. The presence of a fracture line
characteristics of fractured teeth differ and an increase in the proportion of the
volume of the natural tooth crown occupied
Presence of a from those of sound teeth and, thus, are
by the restoration substantially increased
fracture line in associated with fracture. the odds of fracture (P < .001). Additional
The existing literature identifies a
the enamel and risk indicators were unique to the case
number of putative risk indicators for
proportional fracture. The most frequently men- subject–control subject model, including
volume of tioned risk indicator is a large intra- subject age and other measures related to
the relative size of the restoration or to loss
the restoration coronal restoration,1-6 with fewer than
of dentinal support. Neither patient behav-
were strongly 10 percent of fractures occurring in
1,4-6 iors such as clenching, grinding and biting
associated with teeth without restorations. The
hard objects nor occlusal characteristics
effects of restorations are thought to be
the risk of cusp such as guidance, cusp anatomy and gen-
associated with a reduced amount of
fracture. dentin supporting the cusps of a eral wear patterns were strong predictors of
restored tooth.5,6 Larson and colleagues7 fracture risk.
reported that proportional isthmus Conclusions. Among posterior teeth
width is one measure of “lost dentinal support” asso- with restorations, two clinical features were
ciated with in vitro fracture resistance. Numerous strongly associated with the risk of cusp
studies also have reported an association between fracture: presence of a fracture line in the
endodontic treatment without subsequent cusp protec- enamel and proportional volume of the
tion and tooth fracture,4,8-12 although it may be more fre- restoration.
quently associated with incomplete and complete ver- Clinical Implications. Dentists
tical fractures than with complete cusp fractures.4,8,10,11 assessing the risk of fracture should con-
Other risk indicators for complete cusp fracture that sider a detectable fracture line or a high
have been mentioned in the literature include bruxism ratio of restoration-to-total-crown volume as
and worn teeth2,5,13,14; steep cuspal anatomy2,4,5; traumatic important indicators of elevated risk.
occlusal relationships2,5; isolated tooth position5; sharp

JADA, Vol. 135, July 2004 883

Copyright ©2004 American Dental Association. All rights reserved.
R E S E A R C H

cavity preparation internal line angles15; and spe- lighting or drying of the teeth.
cific habits, foods and sweets.16 Specific age pat- For case subjects (that is, those with a frac-
terns have not been strongly associated with com- tured tooth), the dental auxiliaries collected
plete cusp fracture,17 with various studies descriptive clinical data for the fractured (case)
reporting patient age distributions reflecting both tooth and for a comparison tooth, which was the
younger5 and middle-to-older ages.4,18 However, contralateral tooth or its acceptable substitute. A
because no control subjects were included in these contiguous first molar and first premolar could be
studies, it is likely that each distribution was substituted for a missing or crowned second
influenced by the age distribution of patients molar and second premolar, and vice versa.
from whom the data were collected. We recruited control subjects from the same
The lack of definitive information in the litera- patient population as case subjects and enrolled
ture regarding risk indicators for cusp fracture is them at approximately the same rate. We
reflected in the variation among dentists with selected the tooth type (for example, molar, pre-
regard to their ratings of importance of various molar) for which clinical data would be collected
risk indicators for fracture and their assessments to maintain a distribution approximately similar
of the relative risk of fracture of individual teeth.19 to that for case teeth enrolled to date at the same
This study, the first to our knowledge to include a site. No matching was performed between case
direct comparison group, quantifies the odds of and control subjects. We collected clinical data for
complete cusp fracture associated with specific between two and four teeth for each control sub-
tooth and patient-level clinical indi- ject (that is, all maxillary or
cators, as well as patient-level mandibular premolars or molars) to
behaviors and extraoral character- The lack of definitive maximize the statistical power of
istics. Thus, it provides information information is the analysis.
that should help dentists more A minimum of two teeth of the
reflected in the
accurately assess the risk of frac- selected tooth type had to be pres-
ture of one or more cusps in indi- variation among ent and uncrowned for the control
vidual teeth. dentists with regard subject to be eligible for enrollment.
to their ratings of The clinical data collected for case
SUBJECTS AND METHODS
importance of various and comparison teeth included the
Data collection. We conducted a risk indicators for presence or absence of mobility,
case-control study at the same site Class V restorations, cervical
fracture.
as our previous study of the inci- defects, craze lines, tactilely
dence of fractured teeth,1 a large detectable fracture lines, subsurface
dental group practice in Portland, discoloration and endodontic access
Ore. For one year, we asked patients who were preparations. In addition, we noted the restora-
found to have a fractured tooth to participate in tive material and restored surfaces, whether the
the study after informing them of the study’s pur- tooth supported a partial denture, and canine or
pose. We also asked similar numbers of patients group guidance for left and right function.
who did not have a fractured tooth to participate We categorized restorations as being complex if
as control subjects. All informed consent pro- they involved two or more surfaces, with one sur-
cedures, as well as the study procedures and face being a proximal surface; we categorized all
design, were approved by institutional review other restorations as simple. In most, but not all,
boards at the investigators’ institution and at the instances, the clinical examination took place
study site. before any treatment of the fractured tooth. When
Participation consisted of permitting us to col- treatment had been administered, we sought
lect clinical data, alginate impressions and information describing the pre-existing restora-
occlusal records; completing a questionnaire; and tion from recent radiographs and clinicians’
approving the use of data and radiographs from comments.
the clinical record. Two dental auxiliaries con- All subjects completed a 14-item questionnaire
ducted the clinical examinations; they had been that elicited demographic data, as well as infor-
trained and standardized before the start of data mation about behaviors, experiences and symp-
collection. The examinations were performed in toms that might be associated with tooth fracture.
dental operatories without magnification, special Behaviors included bruxism, clenching, chewing

884 JADA, Vol. 135, July 2004

Copyright ©2004 American Dental Association. All rights reserved.
 R E S E A R C H

ice or other hard foods, and biting or holding fractures confirmed our earlier findings that
objects in the mouth. Experiences included pre- unrestored teeth suffered fractures only infre-
vious tooth fracture, recent blows to the face and quently,1 we examined risk indicators for fracture
being warned about possible tooth fracture by a among only those teeth with restorations. Simi-
dentist. Symptoms included pain while chewing larly, because we wished to investigate risk indi-
and sensitivity to hot and cold. cators for fracture associated with teeth that oth-
One of four dental students who had been erwise would not require intervention, we
trained and standardized before data collection excluded teeth with caries. For these reasons, we
analyzed the casts, bite records and radiographs. limited the analyses to subjects with fractured
The students made all measurements at a central teeth that were restored before the fracture
location to record additional characteristics of occurred and did not have a carious lesion asso-
subjects and teeth. For subjects, the dental stu- ciated with the fracture.
dents examined the general extent of faceting Of 249 case teeth for which data were collected,
from the bite record to characterize occlusal wear three had not been restored. When carious teeth
as light, moderate or heavy. For teeth, the stu- were eliminated, one of 201 had not been
dents analyzed radiographs to determine the restored. To maintain the essential premise in
presence of pins, an opposing crown and case-control studies that control observations
endodontic fillings. If an endodontic filling was have had the same opportunity for exposure to
present, the student combined this information potential risk indicators, we included only
with the endodontic-access observa- restored noncarious comparison
tion made clinically to categorize a teeth in control subjects in the anal-
tooth as having had any endodontic The authors yses (that is, 749 control teeth in
treatment. The students examined performed bivariate 252 control subjects).
casts to determine the presence of Data analysis. We conducted
analyses of each
capped cusps (that is, replaced or two separate analyses. One analysis
covered by a restoration), and noted tooth and subject compared the characteristics of case
teeth with one or more unrestored characteristic subjects and case teeth (that is,
cusps that had cuspal inclines of according to those with fractures) with charac-
less than 30 degrees (that is, “flat”). fracture status. teristics of control subjects and com-
The students determined the rel- parison teeth. The other analysis
ative volume proportion, or RVP, of compared characteristics of case
a restoration by calculating the area and comparison teeth in case sub-
of a restoration as a proportion of the coronal area jects. The first analysis represented the principal
in two dimensions (that is, occlusal [from the cast] comparison, with the second comparison of case
and cross-sectional [from the radiograph]) and and comparison teeth in case subjects regarded as
then multiplying these two proportions together. confirmatory for tooth characteristics. We per-
Occlusal areas were determined electronically formed bivariate analyses of each tooth and sub-
from scanned images of the cast and radiograph. ject characteristic according to fracture status.
For the purpose of calculating occlusal surface We determined statistical significance using
areas, students assumed that missing cusps Cochran-Mantel-Haenszel tests and t tests using
reflected standard anatomical relationships. In SUDAAN20 software, which adjusts for the corre-
addition, they used the scanned occlusal images to lated nature of the data (teeth within subjects).
determine the isthmus width (measured on a line For the principal analysis, we used SUDAAN
between cusps) and the mean and maximum logistic regression to identify characteristics asso-
isthmus widths if two were present. ciated with fracture. Because of the number of
Similarly, for each tooth with a restoration, the observations available and the number of inde-
students calculated the mean distance and the pendent variables we wished to test, we were not
shortest distance between the restoration margin able to develop models for different types of pos-
and all cusp tips. From these measures, they cal- terior teeth or for specific cusps within a tooth
culated the proportion of the mean intercuspal type. Instead, we included control variables for
distance represented by the restoration isthmus tooth type in the analyses.
or isthmuses. We used backward-selection methods, first
Subjects. Because the raw distributions of entering all categorical variables and then elimi-

JADA, Vol. 135, July 2004 885

Copyright ©2004 American Dental Association. All rights reserved.
R E S E A R C H

TABLE 1

DEMOGRAPHIC CHARACTERISTICS OF CASE AND CONTROL SUBJECTS.

SUBJECTS SEX (PERCENTAGE) RACE (PERCENTAGE) MEAN (SD*) AGE (YEARS)

Male Female White Asian Other

Cases 47 53 91 4 5 49.6 (11.3)

(n = 200)

Controls 35 65 87 5 8 39.9 (13.5)

(n = 252)

P value .02 .41 .0001

* SD: Standard deviation.

TABLE 2

COMPARISONS OF FRACTURE RISK–RELATED BEHAVIORS AND

SYMPTOMS IN CASE AND CONTROL SUBJECTS.
BEHAVIOR OR PERCENTAGE OF CASE SUBJECTS (n = 200)* PERCENTAGE OF CONTROL SUBJECTS (n = 252)* P VALUE
SYMPTOM
Almost Never Occasionally Frequently Almost Never Occasionally Frequently

Grind When 61 31 9 65 24 11 .16

Asleep

Clench or Grit 41 43 17 44 44 12 .43

Teeth

Chew Ice 63 28 10 64 27 19 .94

Chew Hard 21 62 18 24 67 9 .03

Foods

Bite or Hold 73 24 3 68 29 3 .48

Objects

Pain While 67 29 5 78 19 2 .02

Chewing

Sensitivity to 55 38 8 63 35 3 .02
Hot

Sensitivity to 36 55 10 39 51 10 .78
Cold

* Rows may not total 100 percent because of rounding.

nating those that showed weak or no relation- its analysis. We also eliminated one highly
ships with fracture in a series of small groups. We skewed variable from the analysis (presence of
then added all continuous variables and, again, pins), because few restorations in either group
eliminated those with the weakest relationships. contained pins. In reporting the results of the
Because there were multiple opportunities for co- regression analyses, for some risk indicators, we
linearity (strong relationships between risk indi- present the odds ratio associated with a clinically
cators), we tested all groups of eliminated vari- meaningful difference in the value of the indi-
ables to ensure that they did not contribute to the cator, rather than the odds ratio associated with
explanatory power of the model. the difference between the minimum and the
Before conducting the principal analysis, we maximum values in the analysis.
truncated the range for some continuous vari- By necessity, the within-case-subject analysis
ables to eliminate extreme values, and converted could compare only tooth-level characteristics
one continuous variable to a four-category vari- between case (fractured) and comparison teeth.
able (intercuspal restoration width) to facilitate We used conditional logistic regression (SAS

886 JADA, Vol. 135, July 2004

Copyright ©2004 American Dental Association. All rights reserved.
 R E S E A R C H

TABLE 3

COMPARISONS OF RISK INDICATORS IN CASE AND CONTROL SUBJECTS.

SUBJECT CHARACTERISTIC PERCENTAGE OF CASE SUBJECTS PERCENTAGE OF CONTROL P VALUE
(n = 200) SUBJECTS
(n = 252)

Canine Guidance 71* 61* .02

Previously Fractured Tooth 75 42 < .0001

Recent Blow to Face 2 4 .12

Dentist Warned of Possible 32 17 .0002

Fracture

Light General Wear 17 25* .05

Proportion of Posterior 79 66 < .0001

Teeth With Restorations

Proportion of Posterior 10 7 .01

Teeth With Crowns

* Data were unavailable for one subject.

PHREG procedure, SAS Institute, Cary, N.C.), the putative at-risk condition). In addition, case
entering all available variables for teeth of sub- subjects exhibited canine guidance on the affected
jects in whom the case and comparison teeth had side significantly more frequently than did con-
been restored before the fracture occurred. trol subjects, and more case teeth than control
teeth were classified as mobile.
RESULTS Table 5 (page 889) presents the results of the
Table 1 shows the demographic characteristics of principal and confirmatory regression analyses.
the case and control subjects. Case subjects were For the principal analysis of case and control sub-
significantly older than control subjects, and were jects, the strongest categorical risk indicator was
more likely to be male. The two groups had sim- the presence of a tactilely detectable fracture line,
ilar racial compositions. which raised the odds of fracture by about 75-fold.
Table 2 presents the distributions of fracture Presence of a Class V restoration and a notation
risk–related symptoms and behaviors for case and about fracture risk in the subject’s dental record
control subjects. The distributions differed signifi- were each associated with about sevenfold
cantly for three symptoms and one behavior. increases in the odds of fracture.
Larger percentages of case subjects reported expe- The RVP was a strong risk indicator, with a 10
riencing tooth sensitivity to hot foods and liquids percent increase in the volume proportion of a
frequently and experiencing occasional or frequent restoration associated with a sixfold increase in
pain while chewing. Larger percentages of case the odds of fracture. Indication of an elevated fre-
subjects also reported chewing hard foods fre- quency of pain on chewing (occasionally compared
quently. However, reports of bruxism and with never, or frequently compared with occasion-
clenching behaviors did not differ significantly ally) increased the odds of fracture by about 70
between case and control subjects. Similarly, the percent. A one-category increase in intercuspal
behaviors of chewing ice and biting or holding restoration width (from < 30 percent to 30 to 39
hard objects in the mouth were not significantly percent; from 30 to 39 percent to 40 to 49 percent;
different between case and control subjects. or from 40 to 49 percent to ≥ 50 percent) in-
Tables 3 and 4 show the distributions of clin- creased the odds of fracture by 16 percent. One
ical risk indicators for case and control subjects capped premolar or two capped molar cusps were
and for teeth, respectively. More than two-thirds associated with a more-than-twofold reduction in
of these risk indicators reflected statistically sig- the odds of fracture, while a 10-year increase in
nificant differences between case and control sub- the patient’s age increased the odds of fracture by
jects and between teeth, and most of these differ- about 70 percent.
ences were in the expected direction (that is, the The remaining risk indicators were paradoxical
case subjects or teeth were more likely to exhibit in their effect. A recent blow to the face was asso-

JADA, Vol. 135, July 2004 887

Copyright ©2004 American Dental Association. All rights reserved.
R E S E A R C H

TABLE 4

COMPARISONS OF RISK INDICATORS IN CASE AND CONTROL TEETH.

TOOTH CHARACTERISTIC PERCENTAGE OF n* PERCENTAGE OF n* P VALUE
CASE TEETH WITH CONTROL TEETH WITH
CHARACTERISTIC CHARACTERISTIC

General

Notation of fracture risk in 7 200 1 743 .002

record

Mobility 3 200 <1 749 .02

Endodontic treatment 6 200 2 749 .02

Removable prosthetic 3 200 2 749 .83

abutment

Flat cusp angulation 40 200 44 749 .21

Opposing tooth is crowned 18 200 11 749 .01

Related to Dentinal Support

Subsurface discoloration 43 200 20 749 < .0001

Restoration involves 87 199 53 745 < .0001

proximal surface

Class V restoration 23 200 5 749 < .0001

Cervical defect 6 200 1 749 .01

Fracture line 36 200 1 749 < .0001

Craze lines 26 200 24 749 .99

One or more cusps covered 13 128 9 746 .17

Proportion of cusps covered 5 128 4 746 .30

Pins present 0 199 1 749 .06

Occlusal restoration area 42 124 27 718 < .0001

Cross-sectional restoration 66 130 43 700 < .0001

area

Relative volume 29 123 14 688 < .0001

proportion

Minimum distance from 3.2 128 3.2 714 .99

restoration to cusp tip (mm†)

Mean distance from 4.2 128 4.5 714 .053

restoration to cusp tip (mm)

Widest isthmus width (mm) 7.9 119 6.3 648 < .0001

Mean isthmus width (mm) 7.3 119 5.8 648 < .0001

Intercuspal restoration width 46 133 40 648 < .0001

* Because of missing data for some variables, the total number of teeth varies.
† mm: Millimeter.

ciated with an almost tenfold reduction in the the odds of fracture by 29 percent.
odds of fracture. A 1-millimeter increase in the The confirmatory—or within-case-subject—
shortest distance from the restoration margin to model identified two tooth-level risk indicators as
the cusp tip increased the odds of fracture by 31 significant predictors of fracture: presence of a
percent, and a 10 percent increase in the propor- fracture line and RVP. Both of these indicators
tion of posterior teeth with restorations decreased also were strongly associated with the risk of frac-

888 JADA, Vol. 135, July 2004

Copyright ©2004 American Dental Association. All rights reserved.
 R E S E A R C H

TABLE 5

LOGISTIC REGRESSION MODELS FOR CASE AND CONTROL SUBJECTS

AND TEETH.
RISK INDICATOR ODDS RATIO 95% CONFIDENCE PERCENTAGE (OR MEAN)
INTERVAL IN CASE GROUP

Subjects

Increasing Risk

Fracture line 75.19 25.1 to 225.3 36

Class V restoration 6.89 2.3 to 20.3 23

Notation of fracture risk in 6.69 1.2 to 39.0 7

dental record

RVP* (10% increase) 5.84 3.0 to 11.3 29

Pain when chewing (frequently 1.71 1.6 to 5.5 5 (frequently);

vs. occasionally or occasionally 29 (occasionally)
vs. never)

Age (10-year increase) 1.65 1.1 to 2.4 49.6 (mean, years)

Intercuspal restoration 1.82 1.4 to 2.9 46

proportion (one-category
increase)

Distance from restoration to 1.31 1.2 to 1.5 4.3 (mean, mm†)

cusp tip (1-mm increase)

Decreasing Risk

Posterior teeth with restoration 0.71 0.59 to 0.85 80

(10% increase)

Proportion of cusps covered 0.42 < 0.01 to 0.3 13

(addition of one premolar/two
molar cusps)

Recent blow to face 0.08 0.02 to 0.35 2

Teeth in Case Subjects

Fracture line 7.6‡ 2.9 to 19.5 36

RVP (10% increase) 455.4‡ 13.0 to 1.6 × 105 29

* RVP: Relative volume proportion.

† mm: Millimeter.
‡ Hazard ratio (rather than odds ratio).

ture in the principal analysis that used both case study of the importance of the relative size of the
and control subjects. restoration in determining the risk of fracture,
and the importance of tactilely detectable enamel
DISCUSSION fracture lines in signaling enhanced risk.
The two regression models in our study (com- Relative volume proportion. The RVP is a
paring case and control subjects and case and measure of relative—as opposed to absolute—
comparison teeth in case subjects) had in common restoration size. Thus, it can be considered an
two risk indicators for cusp fracture: the presence inverse measure of the amount of original tooth
of a fracture line and the RVP of the restorations. material (dentin and enamel) remaining. The risk
Both of these risk indicators have face validity, indicators available for inclusion in the model
and large intracoronal restorations in particular consisted of several measures related to absolute
have been cited frequently as both a risk indi- and relative restoration size, including mean and
cator and a risk factor.1-7 However, this finding maximum isthmus width measures, mean and
provides the first confirmation from a controlled minimum margin-to-cusp-tip measures, mean

JADA, Vol. 135, July 2004 889

Copyright ©2004 American Dental Association. All rights reserved.
R E S E A R C H

intercuspal restoration proportion, subsurface differences in cavity design over time of both prin-
discoloration, and presence and proportion of cipal restorations and Class V restorations (with
capped cusps. respect to sharp internal angles and unsupported
RVP emerged as the most powerful of these enamel), as well as differences in expansion char-
risk indicators in both multivariate analyses; this acteristics of restoration materials. Although the
supports the assumption that propensity for cusp available literature does not offer strong evidence
fracture is best determined via a three- of age as a risk indicator for cusp fracture,17 pre-
dimensional assessment of the total amount of vious studies, as noted above, did not include con-
dentinal support remaining for these cusps,5,6 trol groups. That we found an association
rather than a two-dimensional assessment of between a notation of fracture risk in patients’
dentinal support (such as that provided by atten- records and cusp fracture suggests that dentists’
tion to isthmus width or margin-to-cusp measure- subjective impressions of the fracture risk of indi-
ments). The vast majority of restorations in the vidual teeth have some validity.
case teeth were amalgams; only three principal Two additional subject-level characteristics—
restorations were resin-based composites. The having a larger proportion of posterior teeth with
available data did not permit us to determine restorations and having experienced a recent
when case or control restorations had been blow to the face—were associated with a
placed, so we were unable to determine if restora- decreased risk of fracture. Explanations for these
tion age is an independent risk indicator. two associations are less apparent. The protective
Enamel cracks in posterior teeth have long effect of facial blows, which were reported by only
been thought to be a possible pre- 2 percent of case subjects and 4
cursor of cusp fracture. A recent percent of control subjects, may be
observational study showed associa- Increased fracture an anomaly related to the skewed
tions between such cracks and the risk was associated distributions for this measure.
presence of restorations and excur- with increased age, Although the same explanation
sive interferences.21 Because a case- frequent pain while could apply to the proportion of
control study design does not posterior teeth restored, it is less
chewing and a
permit assessment of case teeth credible. The bivariate relationship
before they have fractured, we dentist’s note in the is opposite from the expected direc-
cannot be sure that the fracture patient’s dental tion, with case subjects exhibiting a
lines were evident before the cusp record about the risk greater proportion of restored pos-
fractured. For this reason, we tested of fracture. terior teeth than control subjects.
preliminary principal models that Thus, it is more likely that the
excluded the fracture-line variable, appearance of this risk indicator
and found that the exclusion did not materially signals an overspecification of the effects of other
change the risk indicators included in the model risk indicators in the model; in effect, the model
or their relative strengths of association with has too many measures of the same basic phe-
fracture. This suggests that this risk indicator nomenon, which is dentinal support.
was not masking, or serving as a surrogate for, Tooth-level characteristics. The remaining
other risk indicators. three variables in the model addressed tooth-level
Subject-level characteristics. The principal characteristics: the presence of a Class V restora-
model included several additional risk indicators, tion, intercuspal restoration proportion and
most of which are subject-level characteristics. shortest distance from restoration to cusp tip. All
Increased fracture risk was associated with of these variables assess aspects of the extent of
increased age, frequent pain while chewing and a dentinal support, and their inclusion in addition
dentist’s note in the patient’s dental record about to the RVP indicator attests to the importance of
the risk of fracture. Pain may be a function of dentinal support when assessing fracture risk.
fracture lines, which also were associated with However, the distance measure indicates that
increased fracture risk, while age may be a func- when the shortest distance from margin-to-cusp
tion of decreasing dentin resiliency or cumulative tip is increased, the tooth is at increased risk of
nontraumatic occlusal stress. developing a fracture. This finding is exactly
Alternatively, patient age may be a surrogate opposite from the expected relationship (that is,
for restoration age, which, in turn, may reflect increasing the distance from margin-to-cusp tip

890 JADA, Vol. 135, July 2004

Copyright ©2004 American Dental Association. All rights reserved.
 R E S E A R C H

reduces the fracture risk), which was apparent in used markers of inadequate dentinal support,
the bivariate analysis. Again, the relatively small such as the intercuspal width proportion, have
effect (30 percent reduction in the odds of frac- some validity; however, of perhaps more impor-
ture) suggests that this risk indicator entered the tance is the observation that a quantitative three-
model as a correction for overspecification of other dimensional measure of dentinal support, the
indicators related to dentinal support. RVP, appears to be more strongly associated with
Confirmatory regression model. The con- the risk of fracture. Thus, when assessing risk,
firmatory regression model included only the RVP clinicians should consider the depth as well as the
and fracture-line risk indicators. The reduced width of restorations.
number of indicators is a function of the smaller Furthermore, the study results clearly show
sample size of comparison group teeth (n = 200 that fracture lines that are detectable with an
rather than n = 752) and the lack of subject-level explorer should be considered strong indicators of
risk indicators for entry into the model. With a elevated risk of fracture. Finally, the study
smaller number of comparison teeth in the results suggest that age is related to the likeli-
analysis, some differences between case and hood of fracture, but we were unable to show that
comparison teeth may not be statistically signifi- patient behaviors and patient-level occlusal char-
cant. Because case and comparison teeth in these acteristics are useful as risk indicators. It may be
analyses come from the same individual and are that these behaviors and characteristics are not
paired, subject characteristics cannot differ related to risk, but it is also possible that the
between cases and controls. study did not have sufficient power to detect them
We should note that several because they are influential in
patient behaviors anecdotally only a small proportion of all frac-
assumed to place patients at higher The study results ture incidents.
risk of developing cusp fracture, such confirm the
as clenching, grinding, chewing ice CONCLUSION
relationship between
and biting or holding hard objects, In general, case-control studies
fracture and
showed no associations with fracture cannot definitively identify causes
in this study. Also, the three poten- dentinal support. of a condition, and some of the risk
tial risk indicators related to occlu- indicators in our models clearly
sion that we evaluated in this study are not causal (such as the pres-
were not strongly associated with fracture. Canine ence of a fracture line or a notation by a dentist
guidance, assumed to be protective, was more fre- addressing fracture risk). However, the majority
quent in case subjects than in control subjects in of the indicators in our case-control model do
bivariate analyses. address the concept of dentinal support, which
Our assessment of the extent of tooth wear clearly is predisposing for fracture, and may well
showed no differences in the proportions of case be more important to patient treatment than are
and control subjects with light wear, and the pro- the actual causes of fracture.
portion of case teeth with flat cusp angulation, The situation is akin to a tree that has been
which also is assumed to be protective, was not partially gnawed by a beaver, and later is felled
significantly different from the proportion of com- by a windstorm. The proximate cause, the wind-
parison teeth with flat cusp angulation. However, storm, is of less clinical significance regarding
the occlusal characteristic perhaps most often prevention than is the predisposing factor of loss
associated anecdotally with cusp fracture—trau- of support caused by the gnawing beaver. Recog-
matic occlusion of the involved cusp—could not be nizing the risk associated with loss of dentinal
assessed in this study because the fractured cusp support may be more important to the ultimate
was no longer in occlusion. prevention of cusp fracture than is identifying
The results of this study should be useful to which of a variety of possible proximate causes is
clinicians when assessing a tooth’s risk of fracture most likely to operate in a given patient. ■
because they shed light on which risk indicators Dr. Bader is a research professor, Department of Operative Den-
are valid markers of elevated risk. The relation- tistry, School of Dentistry, CB# 7450, University of North Carolina,
ship between fracture and dentinal support, long Chapel Hill, N.C. 27599-7450, e-mail “[email protected]”. Address
reprint requests to Dr. Bader.
noted in uncontrolled studies, has been con-
firmed. The results demonstrate that commonly Dr. Shugars is a professor, Department of Operative Dentistry, Uni-

JADA, Vol. 135, July 2004 891

Copyright ©2004 American Dental Association. All rights reserved.
R E S E A R C H

versity of North Carolina at Chapel Hill. 8. Schweitzer JL, Gutmann J, Bliss R. Odontiatrogenic tooth frac-
ture. Int Endod J 1989;22(2):64-74.
At the time the data were collected for this study, Dr. Martin was 9. Gher M, Dunlap R, Anderson M, Kuhl L. Clinical survey of frac-
affiliated with Kaiser Permamente Center for Health Research, Port- tured teeth. JADA 1987;114(2):174-7.
land, Ore. She now is in private practice in Canby, Ore. 10. Hansen E. In vivo cusp fracture of endodontically treated premo-
lars restored with MOD amalgam or MOD resin fillings. Dent Mater
This study was supported by NIDCR grant RO1-DE12635. 1988;4(4):169-73.
11. Pietrokovski J, Lantzman E. Complicated crown fractures in
The authors thank OraLee Olsen, Donna Clark and Chris Stermon adults. J Prosthet Dent 1973;30:801-6.
for their efforts in clinical data collection in Oregon, as well as the staff 12. Hansen E, Asmussen E, Christiansen N. In vivo fractures of
and patients of the participating dental offices. In North Carolina, the endodontically treated posterior teeth restored with amalgam. Endod
authors thank John Sturdevant for substantial original contributions Dent Traumatol 1990;6(2):49-55.
to the data collection procedures, Paula Henao-Varner, Shamshia 13. Cameron CE. Cracked-tooth syndrome. JADA 1964;68:405-11.
Shafi, Corey Wilson, Paulette Pauley, Eric Burgin, Samuel Emrich and 14. Cameron C. The cracked-tooth syndrome: additional findings.
Eric Kerr for additional data collection, and Gary Koch and Zhiying JADA 1976;93:971-5.
Pan for statistical support. 15. Vale W. Cavity preparation and further thoughts on high speed.
Br Dent J 1959;107:333-40.
1. Bader J, Martin J, Shugars D. Incidence rates for complete cusp 16. Patel D, Burke F. Fractures of posterior teeth: a review and
fracture. Community Dent Oral Epidemiol 2001;29:346-53. analysis of associated factors. Prim Dent Care 1995;2(1):6-10.
2. Braly B, Maxwell E. Potential for tooth fracture in restorative den- 17. Ellis SG, McFarlane TV, McCord JF. Influence of patient age on
tistry. J Prosthet Dent 1981;45:411-4. the nature of tooth fracture. J Prosthet Dent 1999;82:226-30.
3. Snyder D. The cracked-tooth syndrome and fractured posterior 18. Talim S, Gohil K. Management of coronal fractures of permanent
cusp. Oral Surg 1976;41:698-704. posterior teeth. J Prosthet Dent 1974;31(2):172-8.
4. Lagouvardos P, Sourai P, Douvitsas G. Coronal fractures in pos- 19. Bader J, Shugars D, Roberson T. Using crowns to prevent tooth
terior teeth. Oper Dent 1989;14(1):28-32. fracture. Community Dent Oral Epidemiol 1996;24(1):47-51.
5. Eakle W, Maxwell E, Braly B. Fractures of posterior teeth in 20. Shah B, Barnwell B, Hung P, LaVange L. SUDAAN user’s
adults. JADA 1986;112:215-8. manual, release 5.50. Research Triangle Park, N.C.: Research Triangle
6. Cavel W, Kelsey W, Blankenau R. An in vivo study of cuspal frac- Institute; 1991.
ture. J Prosthet Dent 1985;53(1):38-42. 21. Ratcliff S, Becker I, Quinn L. Type and incidence of cracks in pos-
7. Larson T, Douglas W, Geistfeld R. Effect of prepared cavities on terior teeth. J Prosthet Dent 2001;86(2):168-72.
the strength of teeth. Oper Dent 1981;6(1):2-5.

892 JADA, Vol. 135, July 2004

Copyright ©2004 American Dental Association. All rights reserved.