Accuracy of open tray implant impressions: an in vitro comparison of stock

versus custom trays

Jason Burns, MSc, BDS, LDS, DGDP,a Richard Palmer, PhD, BDS,b Leslie Howe, BDS,c and
Ron Wilson, PhD, MPhild
Guys Hospital, London, England
Statement of problem. The accuracy of an implant fixture-level impression is affected by the type of impression tray used.
Purpose. The purpose of this in vitro study was to investigate the accuracy of open tray implant impressions
comparing polycarbonate stock impression trays and rigid custom-made impression trays to make implant
fixture-level impressions.
Material and methods. Gold cylinder pairs, splinted by gold bars (reference frameworks) were constructed on
an aluminum typodont. Polyether impressions were made of 2 pairs of Brnemark 3.75-mm diameter fixtures
mounted in an aluminium typodont, with 3 stock impression trays, 3 close-fit custom trays, and 3 spaced custom
impression trays, by use of an open tray technique. The casts produced were assessed for accuracy by attaching the
reference frameworks with alternate single screws and measuring the vertical fit discrepancy of these reference
frameworks to the analogs within the working cast using a traveling microscope. Comparison of gap dimensions
by tray type was performed with a nonparametric Kruskal-Wallis analysis of variance (ANOVA) followed by
pair-wise Mann-Whitney U tests. To adjust for multiple comparisons in the post-ANOVA contrasts (P.02).
Comparison of gap dimensions between anterior and posterior regions were performed with Mann-Whitney U
tests (P .05).
Results. The results showed that the mean fit accuracy, as measured by vertical fit discrepancy, of casts from the
stock trays (23 20 m) were statistically significantly less (P.001) than the spaced custom trays (12 10 m)
or close fit custom trays (11 10 m). The difference in median gap size for analogs with a 20-mm separation
was 10 m.
Conclusion. Within the limits of this in vitro study, rigid custom trays produced significantly more accurate
impressions than the polycarbonate stock trays. The stock trays used in this study could not produce accurate
impressions consistently. For analogs with a 20-mm separation, there was a difference in medians of 10 m in
accuracy between the stock and custom trays. (J Prosthet Dent 2003;89:250-5.)

CLINICAL IMPLICATIONS
This in vitro study demonstrated that rigid custom trays produced significantly more accurate
implant fixture-level impressions, as measured by vertical fit discrepancy, than did the polycarbonate stock trays tested.

any clinicians and authors have addressed the

idea that passive fit of implant prostheses is essential to
long-term treatment success.1-5 Although it is assumed
that a misfitting prosthesis between 2 or more implants
may have a negative effect on the long-term stability of
those implants, evidence to support this theory is lacking.6
The statistical correlation between prosthesis misfit
and marginal bone level changes in maxillary implants
with in vivo measurements has been examined.7 This
human retrospective study found that although none of
the prostheses were passively fitting, no evidence of
bone loss was present even after 5 years. One of the
Private practice, London, England.
Professor of Periodontology and Implant Dentistry.
c
Consultant in Restorative Dentistry.
d
Dental Clinical Research.

conclusions from this study was that there must be a

range of prosthesis misfit tolerated by osseointegrated
implants that allows for long-term stability. Work supporting this theory has found that clinically well-fitting
prostheses still produced a considerable amount of misfit
load but no loss of osseointegration.8 Clinically, it is
difficult to measure gaps less than 60 m.9
Although there is some evidence that prosthesis misfit may not affect osseointegration, there is evidence that
prosthesis misfit is likely to increase the incidence of
mechanical component loosening or fracture.10 The
causes of component failure and loosening are multifactorial, but it must be assumed that prosthesis misfit plays
an important role in complications such as occlusal and
abutment screw loosening and fracture in linked implant
restorations.10-13 Because of these concerns, prosthesis
misfit should be minimized.

250 THE JOURNAL OF PROSTHETIC DENTISTRY

VOLUME 89 NUMBER 3

BURNS ET AL

One of the crucial steps for producing a well-fitting

prosthesis is an accurate impression. Most of the available literature evaluating the accuracy of impression materials and comparing accuracy of stock with custom
trays use measurements relating to prepared teeth.14-19
Studies evaluating implant impressions tend to concentrate on comparing different impression techniques by
use of machined components.20-29 The literature suggests that direct impression copings produce more accurate impressions than reseatable impression coping techniques.23-26 It also seems prudent to use a rigid
elastomeric impression material, such as polyether, because it is rigid and maintains impression copings accurately, is dimensionally stable, has a good resistance to
permanent deformation, has a low strain in compression
(flexibility), and has a high initial shear strength.23
Studies with prepared teeth show that custom trays
produce more consistently accurate impressions compared with stock trays.14,16,19,30 This may be because the
tray rigidity does not allow distortion of the impression.19 However, it has been shown that a different
thickness of elastomeric impression material in an impression tray can reduce impression accuracy.31 Comparison of impression accuracy with stock or custom
trays has not been investigated with implants and machined transfer components.
The aim of this in vitro study therefore was to evaluate whether custom trays produce more consistently accurate implant fixture-level impressions than stock trays,
by use of an open tray technique. It was hypothesized
that flexibility in stock impression trays would lead to
less consistently accurate impressions than rigid custom
trays.

MATERIAL AND METHODS

Gold cylinder pairs splinted by gold bars (reference
frameworks) were constructed on a custom constructed
aluminium typodont and tested for fit accuracy using a
1-screw test32 with the aid of a traveling microscope
(Travimed; J Swift and Son, London, United Kingdom).
Three polyether impressions (Impregum Penta; 3M
Espe Dental AG, Seefeld, Germany) were made with
each of 3 stock impression trays, 3 close-fit custom trays,
and 3 spaced custom impression trays. In total, 27 impressions were made of 2 pairs of Brnemark 3.75-mm
diameter implants (Nobel Biocare AB, Goteborg, Sweden) mounted in an aluminium typodont, with an open
tray technique.
The casts produced were assessed for accuracy by attaching the reference frameworks with alternate single
screws and measuring the vertical fit discrepancy of these
reference frameworks to the analogs within the working
cast with the aid of a traveling microscope.32
MARCH 2003

THE JOURNAL OF PROSTHETIC DENTISTRY

Fig. 1. Block with displaceable bed.

Typodont construction
The study used 4 regular platform implant analogs
(DCA 711-0; Nobel Biocare) mounted in an aluminum
typodont. Two channels were milled into the typodont
to retain a flexible bed of silicone material (Gemini 2
part silicone model duplicating material; Bracon Ltd,
Etchingham, Sussex, United Kingdom) (Fig. 1).
The width between the channels corresponded to the
width between the tray walls of a polycarbonate mandibular stock impression tray (Size 12 Solo tray; Davis
Healthcare Services Ltd, Potters Bar, Hertfordshire, England). This allowed the seating of impression trays on a
displaceable bed that could potentially allow tray distortion under load.

Reference bar construction

The typodont retained 4 implant analogs, 2 in the
approximate region corresponding to the mandibular
canine teeth and 2 in the mandibular right premolar/
molar regions. The anterior pair and posterior pair each
had 20 mm of separation, as measured from the center
of each analog.
Four implant-level gold cylinders (DCA 814-0; Nobel Biocare AB) were placed on the typodont abutments
and splinted with a 2-mm diameter gold bar (DCB
077-0; Nobel Biocare AB) using a laser welding machine (Neolaser L; Girrbach, Pforzheim, Germany) to
provide an anterior and posterior reference framework.
Corresponding abutment screws (DCA 778-0; Nobel
Biocare AB) provided the connection between the reference bars and analogs. The analogs were then glued
with epoxy resin adhesive (RS Components, Corby, England) into the typodont with the reference bars still
attached. The accurate fit of the reference bars to the
glued analogs was verified by use of a traveling microscope. These reference bars (Fig. 2) were used to verify
the accuracy of casts produced from impressions.
251

THE JOURNAL OF PROSTHETIC DENTISTRY

Fig. 2. Reference framework (showing orientation markings

to aid positioning of traveling microscope).

Impression tray design

Three types of impression trays were used: mandibular polycarbonate stock trays (Size 12 Solo trays; Davis
Healthcare Services Ltd); close-fit custom trays constructed with a single sheet of rigid visible light-polymerizing material custom tray material (Palatray LC;
Heraeus Kulzer, Wehreim, Germany) polymerized over
a 3-mm wax spacer covering the typodont; and spaced
custom trays constructed with a single sheet of rigid
visible light-polymerizing material custom tray material
(Palatray LC; Hereus Kulzer) polymerized over a plaster
horseshoe to allow for 10 mm of space. This horseshoe
was cast from an impression of one of the stock trays.
This ensured that internal dimensions of the stock trays
and the spaced custom trays were the same.
Three stock trays were selected. In addition, 3 closefit and 3 spaced custom trays were fabricated, and all the
trays were opened with 4 holes cut above the typodont
analogs. This allowed access to the guide pins when
making impressions using an open tray technique. Each
tray within each set of 3 was tested to ensure uniform
construction by measurement of tray thickness at numerous points on each tray.

Impression protocol
Three impressions were made from each of the 9
impression trays using implant transfer copings (DCA
099-0; Nobel Biocare AB). Polyether impression material (Impregum Penta; 3M Espe Dental AG) was used
and was dispensed using a delivery unit (Pentamix II;
3M Espe Dental AG). A stop clock was used to note the
time to load and level the tray, load the syringe, syringe
the impression material around the copings, seat the tray
on the displaceable bed, and allow for full setting, to
standardize the impression protocol. The material was
allowed to polymerize for twice the manufacturers recommended setting times to allow for room temperature
rather than mouth temperature. A circular piece of steel
weighing 430 gm was used to standardize the seating
252

BURNS ET AL

Fig. 3. Measurement casts.

load on the impression tray for each specimen. It was

necessary to construct a 2.5-mm-thick horseshoe of
polymethyl methacrylate (Perspex; Cox Plastics, Gateshead, United Kingdom) to fit between the weight and
impression copings to ensure the weight was acting just
on the tray and was not deflected by the transfer copings.

Cast production protocol

Once polymerized, 4 regular platform implant analogs were screwed onto the impression copings contained within the set impression. Care was taken to ensure that the impression analogs were fully seated against
the impression copings.
The impressions were cast in batches of 9 within the
manufacturers recommended times. The impressions
were cast in a type IV dental stone (GC Fujirock EP; GC
Europe, Leuven, Belgium) according to the manufacturers instructions. Each cast was then trimmed into an
anterior and posterior pair (Fig. 3). Each cast of each
pair was carefully marked A for anterior and P for posterior, and each pair was randomly designated a number
between 1 and 6. This number corresponded to a particular tray type and number, and this information was noted
on a sheet and subsequently placed in a sealed envelope. All
the casts were poured and prepared by one operator.

Measurement protocol
The subsequent measurements performed on the
casts were without the previously described information;
the envelope was opened only after the measurements
and re-measurements had been completed. This allowed
for a single blind study design. All measurements were
made by one operator.
By virtue of the markings on the reference framework
and cast, it was a straightforward procedure to mount
the anterior or posterior reference bar on the appropriate cast in the correct orientation. One end of the bar
was then attached to the cast with the abutment screw
and tightened to 20 Ncm with the restorative torque
indicator (RTI2035; 3i Implant Innovations, Palm
VOLUME 89 NUMBER 3

BURNS ET AL

THE JOURNAL OF PROSTHETIC DENTISTRY

Fig. 4. Gap measurement points on stone blocks. A, Anterior

analogues; P, posterior analogues; arrows, points at which
veritcal gaps were measured on each specimen; red arrows,
points that were remeasured 3 days later; dotted lines; each
pair of analogues trimed into a single block specimen.

Beach, Fla.). The accuracy of bar fit was then quantified

by measuring the vertical gap between the other cylinder
and analog with a traveling microscope at 3 points on
each analogue (Fig. 4).
The abutment screw was then removed and tightened onto the other gold cylinder at 20 Ncm, and the
procedure was repeated. The same measurement protocol was used for each cast throughout the study.
Comparison of gap dimensions by tray type was performed with a nonparametric Kruskal-Wallis analysis of
variance (ANOVA) followed by pair-wise Mann-Whitney U tests. To adjust for multiple comparisons in the
post-ANOVA contrasts, statistical significance was assumed where P.02. Comparison of gap dimensions
between anterior and posterior regions were performed
with Mann-Whitney U tests, and P.05 was used to
judge statistical significance.
To test the reproducibility of the measurements
made, 2 of the 6 points on each cast were remeasured
3 days later. The casts had the appropriate reference bars
reattached and were remounted and remeasured by use
of the same standard protocol. The same 2 points were
remeasured throughout (Fig. 4). A summary of the protocol used in the main study is shown in Figure 5.

RESULTS
The results are presented in Tables I and II. The
results showed that the mean fit accuracy, as measured
by vertical fit discrepancy, of casts from the stock trays
(23 20 m) were statistically significantly less
(P.001) than the spaced custom trays (12 10 m) or
close-fit custom trays (11 10 m). A significant difference (P.001) between the stock and the custom
impression trays was found at both anterior and posteMARCH 2003

Fig. 5. Summary of main study protocol.

rior sites. However, there was no statistical difference

between the 2 types of custom impression tray (P.5).
The difference in median gap size between stock and
both types of custom trays for analogs with a 20-mm
separation was 10 m.
There were also significant differences (P.05) between anterior and posterior pairs in all tray types, and
the anterior pairs produced smaller vertical fit discrepancies. There was no significant difference in measurements between separate trays of the same description.
Comparisons between sites and between left and right
pairs (the bars were splinted) were not significant.

DISCUSSION
The study showed that custom trays produced more
accurate impressions than stock trays. This finding is in
agreement with a number of other studies with natural
teeth14,16,19,30; however, there are potentially major differences between the natural tooth studies and this
study. In the natural tooth studies, investigators used
closed impression trays that may have been more susceptible to flexion and distortion when seated with impression material. The impression protocol used for implant
fixture-level registration generally uses an open tray,
which may allow impression material subjected to hydrostatic pressure on seating to more readily escape. In
addition, the use of a fixture-level impression technique
(with machined components) only needs to record the
implant head position rather than both its position and
dimensions. Both of these factors would suggest that
253

THE JOURNAL OF PROSTHETIC DENTISTRY

BURNS ET AL

Table I. Distribution of gap measurements by tray type

Gap Value in
micrometers

Stock trays
Spaced custom trays
Close fit custom trays

10

20

30

40

50

6
27
30

39
48
46

32
21
24

7
10
7

8
2

60

70

80

90

100

110

120

Table II. Results showing vertical gap measurements in micrometers Mean, (standard deviation), median and interquartile
range
Tray Type

Anterior implant sites

Posterior implant sites

P value for comparison of anterior

and posterior implants
All implant sites

Stock

Spaced Custom

Close Fit Custom

18
(17)
20
10-40
28
(21)
20
10-40
0.001

9
(8)
10
0-10
14
(11)
10
10-20
0.013

9
(11)
10
0-10
14
(8)
10
10-20
0.001

23
(20)
20
10-30

12
(10)
10
5-30

11
(10)
10
0-20

P value for
comparison of trays*

.001

.001

All Trays

12
(13)
10
10-20
19
16)
20
10-20
0.001

.001

*Kruskal-Wallis ANOVA. Stock trays significantly different from spaced and close fit custom trays at both anterior and posterior sites (P .001), spaced custom
and close fit custom trays not significantly different (P.5). Post-ANOVA contrasts with Mann-Whitney U test.

Mann-Whitney U test

impression distortion was likely to have less effect than in

conventional fixed prosthodontic impressions.
From examination of the data, there was no statistical
difference in gap measurements between the 2 types of
custom tray. This may have been because the tray rigidity
did not allow distortion of the impression. However, it has
been shown that a different thickness of elastomeric impression material in an impression tray can reduce impression accuracy.31 It was postulated in this study that the
impression material could shrink away from the impression
specimen during setting, because of the impression material preferentially adhering to the tray rather than the specimen. This distorting effect would be magnified with
greater bulks of impression material.
In this study, however, there were no statistically significant differences in vertical gap measurements between the spaced and close fit custom impression trays.
This occurred in spite of a much thicker layer of impression material in the spaced tray (10 mm) than in the
close-fit tray (3 mm). It seems that any influence of the
impression material thickness on the spatial positioning
of the transfer copings was negligible.
There were statistically significant differences in accuracy between the anterior analog pairs and posterior
analog pairs in all trays. For stock trays, the increased accu254

racy in the anterior part of the tray may be accounted for

because the tray is thickest in the midline region and is
therefore more rigid. The differences in gap values between
anterior and posterior pairs in the custom trays are more
difficult to explain. It must be noted for all tray types,
however, that in spite of statistical significance, the actual
differences in anterior and posterior values were small and
may have little clinical significance.
The results of this study showed a difference in medians of 10 m in accuracy between the stock and custom trays, as measured by vertical fit discrepancy. This in
itself is probably of minimal clinical significance, especially when it has been shown that, clinically, it is difficult
to measure gaps less than 60 m.9 Over a larger implant
span or with multiple linked fixtures, these measurement
differences could potentially be magnified to a degree
that is clinically significant. However, any extrapolations
of these results into a clinical environment should be
interpreted with caution.
In view of the ideal bench conditions for making
impressions, one would expect to obtain consistently
accurate results with all trays. However, the stock trays
obviously allowed distortion of the impression. Because
the only discernable difference between the stock and
spaced custom trays were differences in rigidity, this
VOLUME 89 NUMBER 3

BURNS ET AL

must be a contributory factor. One theory derived from

studies on natural teeth is that, under load, the stock tray
flexes and the impression distorts. The other theory of
differing uniform impression thickness may occur, but it
did not appear to influence the results of this study. Why
the stock tray distortions did not occur consistently with
the strict protocol used is difficult to explain, but it is clear
is that an identical but rigid tray eliminated this problem.
The results showed that flexible stock trays were statistically more inaccurate. It was possible to make accurate
stock tray impressions, although the accuracy was not as
consistent compared with custom trays. Provided an accurate impression material and good impression protocol are
used, a rigid stock tray may be a valid alternative to custom
trays for implant fixture-level impressions. This would require further investigation. Even though the difference in
medians of 10 m in accuracy between the stock and custom trays, as measured by vertical fit discrepancy, is statistically significant, this difference may be difficult to identify
and measure if extrapolated clinically.

CONCLUSIONS
Within the limits of this in vitro study, it may be
concluded that, as measured by vertical fit discrepancy,
rigid custom close-fit trays and spaced custom trays produce significantly more accurate impressions than flexible polycarbonate stock trays (P.001). Also, for analogs with a 20-mm separation, there was a difference in
medians of 10 m in accuracy between the stock and
custom trays, as measured by vertical fit discrepancy.
REFERENCES
1. Jemt T, Lie A. Accuracy of implant-supported prostheses in the edentulous
jaw: analysis of precision of fit between cast gold-alloy frameworks and
master casts by means of a three-dimensional photogrammetric technique.
Clin Oral Implants Res 1995;6:172-80.
2. Millington ND, Leung T. Inaccurate fit of implant superstructures. Part 1:
Stresses generated on the superstructure relative to the size of fit discrepancy. Int J Prosthodont 1995;8:511-6.
3. Jemt T. In vivo measurements of precision of fit involving implant-supported prostheses in the edentulous jaw. Int J Oral Maxillofac Implants
1996;11:151-8.
4. Riedy SJ, Lang BR, Lang BE. Fit of implant frameworks fabricated by
different techniques. J Prosthet Dent 1997;78:596-604.
5. Jemt T, Lekholm U. Measurements of bone and frame-work deformations
induced by misfit of implant superstructures. A pilot study in rabbits. Clin
Oral Implants Res 1998;9:272-80.
6. Taylor TD, Agar JR, Vogiatzi T. Implant prosthodontics: current perspective and future directions. Int J Oral Maxillofac Implants 2000;15:66-75.
7. Jemt T, Book K. Prosthesis misfit and marginal bone loss in edentulous
implant patients. Int J Oral Maxillofac Implants 1996;11:620-5.
8. Smedberg JI, Nilner K, Rangert B, Svensson SA, Glantz SA. On the influence of superstructure connection on implant preload: a methodological
and clinical study. Clin Oral Implants Res 1996;7:55-63.
9. McLean JW, Frauenhoffer JA. The estimation of cement film thickness by
an in vivo technique. Br Dent J 1971;131:107.
10. Kohavi D. Complications in the tissue integrated prostheses components:
clinical and mechanical evaluation. J Oral Rehabil 1993;20:413-22.
11. Jemt T, Linden B, Lekholm U. Failures and complications in 127 consecutively placed fixed partial prostheses supported by Branemark implants:

MARCH 2003

THE JOURNAL OF PROSTHETIC DENTISTRY

12.
13.

14.
15.

16.
17.

18.

19.

20.
21.
22.

23.
24.
25.

26.

27.
28.

29.
30.

31.
32.

from prosthetic treatment to first annual checkup. Int J Oral Maxillofac

Implants 1992;7:40-4.
Tolman DE, Laney WR. Tissue-integrated prosthesis complications. Int
J Oral Maxillofac Implants 1992;7:477-84.
Kallus T, Bessing C. Loose gold screws frequently occur in full-arch fixed
prostheses supported by osseointegrated implants after 5 years. Int J Oral
Maxillofac Implants 1994;9:169-78.
Gordon GE, Johnson GH, Drennon DG. The effect of tray selection on the
accuracy of elastomeric impression materials. J Prosthet Dent 1990;63:12-5.
Boulton JL, Gage JP, Vincent PF, Basford KE. A laboratory study of dimensional changes for three elastomeric impression materials using custom
and stock trays. Aust Dent J 1996;41:398-404.
Millstein P, Maya A, Segura C. Determining the accuracy of stock and
custom tray impression/casts. J Oral Rehabil 1998;25:645-8.
Saunders WP, Sharkey SW, Smith GM, Taylor WG. Effect of impression
tray design and impression technique upon the accuracy of stone casts
produced from a putty-wash polyvinyl siloxane impression material.
J Dent 1991;19:283-9.
Saunders WP, Sharkey SW, Smith GM, Taylor WG. Effect of impression tray
design upon the accuracy of stone casts produced from a single-phase
medium-bodied polyvinyl siloxane impression material. J Dent 1992;20:18992.
Wassell RW, Ibbetson RJ. The accuracy of polyvinyl siloxane impressions
made with standard and reinforced stock trays. J Prosthet Dent 1991;65:
748-57.
Spector MR, Donovan TE, Nicholls JI. An evaluation of impression techniques for osseointegrated implants. J Prosthet Dent 1990;63:444-7.
Barrett MG, de Rijk WG, Burgess JO. The accuracy of six impression
techniques for osseointegrated implants. J Prosthodont 1993;2:75-82.
Humphries RM, Yaman P, Bloem TJ. The accuracy of implant master casts
constructed from transfer impressions. Int J Oral Maxillofac Implants
1990;5:331-6.
Carr AB. Comparison of impression techniques for a five-implant mandibular model. Int J Oral Maxillofac Implants 1991;6:448-55.
Carr AB. Comparison of impression techniques for a two-implant 15degree divergent model. Int J Oral Maxillofac Implants 1992;7:468-75.
Liou AD, Nicholls JI, Yuodelis RA, Brudvik JS. Accuracy of replacing three
tapered transfer impression copings in two elastomeric impression materials. Int J Prosthodont 1993;6:377-83.
Phillips K, Nicholls JI, Tsun M, Rubenstein JE. The accuracy of three
implant impression techniques: a three dimensional analysis. Int J Oral
Maxillofac Implants 1994;9:533-40.
Assif D, Fenton A, Zarb G, Schmitt A. Comparative accuracy of implant
impression procedures. Int J Periodontics Restorative Dent 1992;12:112-21.
Inturregui JA, Aquilino SA, Ryther JS, Lund PS. Evaluation of three impression techniques for osseointegrated oral implants. J Prosthet Dent 1993;
69:503-9.
Assif D, Marshak B, Schmidt A. Accuracy of implant impression techniques. Int J Oral Maxillofac Implants 1996;11:216-22.
Rueda LJ, Sy-Munoz JT, Naylor WP, Goodacre CJ, Swartz ML. The effect
of using custom or stock trays on the accuracy of gypsum casts. Int J
Prosthodont 1996;9:367-73.
Eames WB, Sieweke JC, Wallace SW, Rogers LB. Elastomeric impression
materials: effect of bulk on accuracy. J Prosthet Dent 1979;41:304-7.
Tan KB, Rubenstein JE, Nicholls JI, Yuodelis RA. Three-dimensional analysis of the casting accuracy of one-piece, osseointegrated implant-retained
prostheses. Int J Prosthodont 1993;6:346-63.

Reprint requests to:

DR JASON BURNS
320 KEW ROAD
RICHMOND
SURREY, TW9 3DU
UNITED KINGDOM
E-MAIL: jason@molars.com
Copyright 2003 by The Editorial Council of The Journal of Prosthetic
Dentistry.
0022-3913/2003/$30.00 0
doi:10.1067/mpr.2003.38

255