Volume 79 • Number 11

Effect of Microthreads and Platform

Switching on Crestal Bone Stress
Levels: A Finite Element Analysis
Jason Schrotenboer,* Yi-Pin Tsao,† Vipul Kinariwala,‡ and Hom-Lay Wang§

Background: The aims of this study were to investigate the

effects of implant microthreads on crestal bone stress com-
pared to a standard smooth implant collar and to analyze
how different abutment diameters influenced the crestal
bone stress level.
Methods: Two-dimensional finite element imaging was
used to create a cross-sectional model of an implant (5-mm

T
he restoration of edentulous areas
platform and 13 mm in length) placed in the premolar region by using dental implants has been
of the mandible. The two tapered implant models consisted well documented and shown to
of one with microthreads at the crestal portion and the other have predictable outcomes.1-3 The lon-
with a smooth neck. The implant model was reverse-engineered gevity of the implants relies primarily on
to resemble a commercially available microthread implant. their stability at placement.1 However,
Abutments of different diameters (4.0 mm: 20% platform early peri-implant bone loss has been
switching; 4.5 mm: 10% platform switching; and 5.0 mm: observed in many implant systems and
standard) were loaded with a force of 100 N at 90 vertical after different surgical approaches.1,3-5
and 15 oblique angles. Finite element analysis was used to In particular, crestal bone loss is the
analyze the stress patterns in bone, especially in the crestal leading symptom of implant failure after
region. osseointegration and the achievement
Results: Upon loading, the microthread implant model had of primary stability. The majority of
29% greater stress (31.61 MPa in oblique and 9.31 MPa in ver- crestal bone loss occurs during the first
tical) at the crestal bone adjacent to the implant than the year of implant function, and it can be as
smooth-neck implant (24.51 and 7.20 MPa, respectively). much as 1.2 mm corono-apically.6 The
When the abutment diameter decreased from 5.0 to 4.5 mm presence of crestal bone is one of the
and then to 4.0 mm, the microthread model showed a reduc- key factors that influences the appear-
tion of stress at the crestal bone level from 6.3% to 5.4% after ance or maintenance of peri-implant
vertical loading and from 4.2% to 3.3% after oblique loading. soft tissue architecture.
The smooth-neck model showed a reduction of stress from There are many suggested causes for
5.6% to 4.9% after vertical loading and from 3.7% to 2.9% after early implant bone loss, two of them be-
oblique loading. ing occlusal overload and implant crest
Conclusions: Microthreads increased crestal stress upon module.5 The crest module of the implant
loading. Reduced abutment diameter (i.e., platform switch- body refers to the transosteal region of
ing) resulted in less stress translated to the crestal bone in the implant that receives the stress from
the microthread and smooth-neck groups. J Periodontol the implant after loading.7 An example of
2008;79:2166-2172. a suspected bone morphology alteration
KEY WORDS due to stress is the apical migration of
crestal bone down to the first thread of
Alveolar bone loss; computer assisted; dental abutment;
many implant systems.1,5,8 It has also
dental implants; numerical analysis.
been hypothesized that the bone loss
may slow down at the first thread when
* School of Dentistry, University of Michigan, Ann Arbor, MI.
the force changes from crest shear force
† Private practice, San Francisco, CA. to the compressive force induced by the
‡ Cranes Software, Troy, MI.
§ Department of Periodontics and Oral Medicine, School of Dentistry, University of thread itself.5 In general, a functional im-
Michigan. plant may encounter many different

doi: 10.1902/jop.2008.080178

2166
J Periodontol • November 2008 Schrotenboer, Tsao, Kinariwala, Wang

forces, such as rotation, shear, and compression. It MATERIALS AND METHODS

was found that the cortical bone layer withstands A two-dimensional finite element model (Fig. 1) of a
compressive force the best.9 completely osseointegrated endosseous titanium im-
Therefore, an implant system should be designed plant in the posterior mandible was created for the
so that it can best distribute stress to bone in a manner purposes of stress analysis in conjunction with design
that supports a restoration in function and encourage alterations. FEA software¶ was used to generate the
osseous attachment. The functionality and longevity model, create the mesh of the individual elements,
of these implant systems rely on the mechanical in- and perform the analysis of the resulting models.
tegrity of the prosthesis and implant10 as well as the The model of the posterior mandible in cross-
ability of peri-implant structures to withstand and pos- section was constructed using measurements and
itively adapt to the applied forces.11 Application of too geometries similar to another study19 with isotropic
much stress can cause bone resorption or even failure material properties (Table 1).20-22 An isotropic mate-
of the implant–bone interface, whereas lack of stress rial is defined as having identical physical properties
may lead to atrophy or even bone loss.12-14 Thus, the in all directions; therefore, only two independent
possibility of mechanical rearrangement leading to an material constants exist. This model consisted of
increase or decrease in peri-implant bone structure thick cortical bone surrounding dense trabecular
quality is highly related to the magnitude and fre- bone, which is classified as type II bone.23 The result-
quency of force applied to it. ing mandibular cross-section was 28 mm in vertical
One design concept is a rough external surface on dimension and 14.1 mm at the greatest horizontal di-
the transosteal portion of an implant fixture. The me- mension. The thickness of the cortical bone section
chanical benefit is an increase in implant–bone con- ranged from 0.595 to 1.515 mm, with the crestal re-
tact available for stress translation.15 One type of gion measuring 1.5 mm.
roughened surface results from the addition of micro- The smooth-neck (control fixture) implant model
threadsi to the neck. A clinical trial16 demonstrated consisted of a restorative platform width of 5 mm, a
possible preservation of crestal bone contact with im- crestal width of 5.5 mm, and a length of 13 mm. The
plant systems using microthreads. dimensions of the implant model selected in this study
A second design strategy in implant systems is a represent one of the most commonly used sizes in clin-
switched platform abutment on an implant fixture. ical practice. The microthread (test fixture) implant
In essence, platform switching is the placement of model was replicated with the exception of micro-
a smaller-diameter abutment relative to the platform threads replacing the smooth-neck portion. The origi-
of the fixture. Primarily associated with the internal- nal diameter was maintained throughout this portion.
ization of the microgap,5 this design has also been Microthreads and main body threads were modeled
suggested as beneficial for crestal bone preserva- as V-shaped projections in which the microthreads
tion.17 Because platform switching involves a change and body threads had a distance of 0.2 and 0.8 mm,
in the implant system design structure, this feature respectively. The smooth and microthread neck por-
may also have a role in stress translation from implant tion was 4.8 mm in height and modeled to be in
to bone. In this study, the term ‘‘platform switching’’ 100% contact with surrounding structures. The 3-mm
refers to a reduced-diameter abutment in relation to tall abutment models consisted of 5-, 4.5-, and 4-mm
the diameter of the fixture platform. diameters.
To investigate the influence of the crestal module, Complete osseointegration at the implant–bone in-
specifically in the platform-switching systems, a com- terface was simulated by combining the nodes of the
puter model was developed to apply finite element implant and bone models. Similar integration of the
analysis (FEA). The use of FEA in implant analy- abutment and implant body was adopted to be a sin-
sis has been widely demonstrated and published.18 gle unit. This eliminated any potential influence from
Because the geometries involved with modeling im- the micromovement between components.
plants and the alveolar process are very complex, The mesh consisting of the implant body, abut-
FEA is considered the most suitable tool for analyzing ment, and bone consisted of 42,159 elements and
them. This type of analysis allows researchers to pre- 41,902 nodes, which was for control and test fixtures
dict stress distribution in the contact area between im- with 5-mm abutments. A two-dimensional (2D) plane
plants and cortical bone as well as around the apex of stress element and h-method discretization were used
the implants in trabecular bone.18 Therefore, the aims for computation and analysis of the model. Boundary
of this study were to investigate the effects of implant conditions were modeled to fix the inferior one-third of
microthreads on crestal bone stress levels compared
to a standard smooth implant collar and to analyze
how different abutment diameters influence crestal i Astra Implant ST, Astra Tech, Mölndal, Sweden.
bone stress levels. ¶ NISA V15, Cranes Software, Troy MI.

2167
Effects of Microthreads and Platform Switching on Crestal Bone Stress Volume 79 • Number 11

the center apical surface of the abutment, through

the implant body, to the peri-implant tissues.
Stress analysis of all models consisted of mapping
von Mises stress patterns upon the application of ver-
tical and oblique loading scenarios. von Mises stress,
a type of applied stress analysis used in FEA, is
measured in MPa. The values used in the comparisons
were located at the most crestal cortical bone, adja-
cent to the implant fixture, in the mandibular model.

RESULTS
In all models, the greatest concentration of stress was
located at the crestal level adjacent to the implant and
at the points of the microthreads and body threads.
The localized crestal stress adjacent to the implant
model decreased drastically within a horizontal dis-
tance of 3 mm. Figures 2 and 3 show the crestal por-
tion of the model adjacent to the smooth neck or
Figure 1. microthreads, respectively.
Implant and surrounding mandibular bone cross-section used for finite Upon oblique loading, the microthread implant
element modeling and analysis. Microthread model shown, with orange model with a control abutment had 29% greater stress
and red sections representing cortical and trabecular bone, respectively. at the crestal bone adjacent to the implant than the
The implant and abutment are shown as blue and green, respectively.
smooth-neck implant. Under vertical load, the micro-
thread model displayed a similar 29% increase in
crestal bone stress. Specifically, von Mises stresses
Table 1. in the microthread model were 31.61 MPa for oblique
Comparison of Material Properties loading and 9.31 MPa for vertical loading; in the
smooth-neck model they were 24.51 and 7.20 MPa,
respectively.
Modulus of Poisson
When the abutment diameter decreased from 5.0 to
Component Rigidity (MPa) Ratio
4.5 mm, and then to 4.0 mm, the microthread model
Titanium (Sakaguchi 117 · 103 0.30 showed a reduction in crestal bone stress levels from
and Borgersen, 199520) 6.3% to 5.4% after vertical loading and from 4.2% to
3.3% after oblique loading, respectively. The stress
Trabecular bone (Borchers 1,370 0.31
and Reichart, 198321) level with the smooth-neck model decreased from
5.6% to 4.9% after vertical loading and from 3.7% to
Cortical bone (Rice et al., 198822) 2,727 0.30 2.9% after oblique loading (Table 2).

DISCUSSION
the mandibular model in all degrees of freedom to The purpose of this study was to use a two-dimen-
minimize stress interferences. sional computer model (e.g., FEA) to analyze the
Model analysis consisted of two groups: smooth effect of microthreads and platform switching on im-
neck (control) and microthread (test), each with three plant crestal bone stress where the greatest stress
abutment diameters (5, 4.5, and 4 mm). The 5-mm was noted. Although 2D and three-dimensional (3D)
abutment represented the control diameter; the 4.5- models have their advantages and disadvantages, this
and 4-mm abutments represented a diameter reduc- study used a 2D model because of ease of manip-
tion of 10% and 20%, respectively. This reduction in ulation, simplicity, efficiency, and cost-effectiveness.
abutment diameter represented the concept of plat- The aim of this study was not to replicate exact in vivo
form switching. stresses, but rather to illustrate a possible difference
Force application was performed in oblique and between a microthread and a smooth-neck implant
vertical conditions using 100 N as a representative counterpart coupled with abutments simulating plat-
masticatory force.11,19,24,25 For oblique loading, a form switching. Because this study was modeled in
force of 100 N was applied at 15 from the vertical 2D, antirotational elements of an implant system were
axis. This translated into ;26.8 N in the horizontal di- not considered crucial and were not implemented into
rection and 96.3 N in the vertical direction. In both the design model, because the force vector was also
cases, the transfer of load was simulated to be from in a 2D plane. The rationale behind the omission of

2168
J Periodontol • November 2008 Schrotenboer, Tsao, Kinariwala, Wang

Figure 2.
Vertical loading stress plots. Top row shows microthread model; bottom row shows smooth-neck model. Key shown in von Mises stress (MPa).

design elements, such as a Morse taper connection bone may have had an effect on the results.19 This
system, was due to the fact that our model was de- study also modeled the bone–implant contact as a con-
signed to have node sharing between the abutment sistent 100% to create a modern model similar to pho-
and implant body. Because of this mating condition, toelastic models, whereas most common bone–implant
there is little significance to the shape of the abutment contact percentages ranged from 30% to 70%. Because
that inserts into the implant body. Micromovement the goal of this study was to investigate the effects
among components of this model was also omitted. on surrounding bone when only two design aspects
Although micromovement plays a vital role in im- (thread pattern and platform size) of an implant system
plant stability, it was determined that micromovement were modified, it was more efficient and minimally com-
would have been consistent between different models plicated to use isotropic values instead of anisotropic
and it would have been redundant to create in each values. Therefore, 2D FEA modeling satisfied the
model. Thus, the simplicity of the modeled abutment, criteria of easily depicting stress differences without
implant, and mandibular cross-section was sufficient using unnecessarily complex geometries that were
to demonstrate the effects of microthreads and plat- viewed to have an insignificant impact on this study.
form switching without adding complicated design A 3D model may demonstrate varying amounts of
parameters not examined in this study. stress in all planes. However, in this 2D model, we
This study assumed isotropic properties for cortical compared stress differences between the abutment
and trabecular bone. The trabecular model used in and implant models because the 2D model only
this study was modeled as a solid, isotropic material showed one plane of the stress pattern. This was much
with no porosity, which is found in in vivo trabecular easier to analyze based on our study objectives.
bone; this is in agreement with a study done by For the analysis portion of this study, it was deter-
Akagawa et al.,26 in which trabecular bone was mod- mined that a vertical and oblique loading model
eled as a solid substance and a more natural porous should be tested. A 15 angle and a loading force of
substance. Using the isotropic property instead of an- 100 N were chosen because this force was shown in
isotropic bone properties for trabecular and cortical other studies11,19,24 to be more comparable to in vivo

2169
Effects of Microthreads and Platform Switching on Crestal Bone Stress Volume 79 • Number 11

Figure 3.
Oblique loading stress plots. Top row shows microthread model; bottom row shows smooth-neck model. Key shown in von Mises stress (MPa).

Table 2. the actual vector of force can vary among individ-

uals.10,27 Because of the constraint on the mandibular
Results from FEA Analysis for Oblique portion of the model, only the inferior one-third of the
and Vertical Loading Conditions mandible was surrounded by a fixed constraint to min-
imize any interference during analysis.
Crestal Stress Adjacent to the Implant Neck (MPa) Data from this study showed that the microthread
Abutment diameter 4.0 mm 4.5 mm 5.0 mm implant had 29% greater stress at the crestal bone ad-
jacent to the implant than the smooth-neck implant in
Vertical loading oblique and vertical load. This result was proportion-
MT 8.26 8.72 9.31 ally similar to another study19 that used similar angles
SN 6.48 6.80 7.20 and load applications. The range (6.48 to 24.51 MPa)
Oblique loading of cortical von Mises stress seen in our study may be
MT 29.29 30.30 31.61 slightly higher than in the in vivo condition and can
SN 22.93 23.62 24.51 be attributed to the linear FEA modeling used in
MT = microthread model; SN = smooth-neck model. the study. Thus, an increase or decrease in force ap-
plication would have respective effects on the data
produced. Nonetheless, based on histologic examina-
mastication and is a biologically feasible action that tion and FEA results, previous studies14,27 showed a
can be performed on an implant in vivo.24,25 Also, stress equivalent to 1.6 MPa was sufficient to avoid
the application of a horizontal vector creates the most crestal bone loss from disuse atrophy in the canine
shear stress in cortical bone and was shown to be mandibular premolar region. On the other end, 60
the component of force best avoided for implant MPa was regarded as a threshold strain value above
success.18 To augment the oblique condition, an ad- which bone failed to heal after fatigue.28 Although
ditional model, with vertical loading of 100 N, was cre- our data were not modeled to be in complete compli-
ated. Although these forces and angle represent ance with an in vivo model, the design model used in
possible applications of force to a dental implant, this study is clinically supported by these findings. Lee

2170
J Periodontol • November 2008 Schrotenboer, Tsao, Kinariwala, Wang

et al.29 demonstrated in human patients that there the company that provided the software to generate
may be maintenance of marginal bone loss in micro- the finite element model. Drs. Schrotenboer, Tsao,
thread implants over smooth-neck systems. The and Wang report no conflicts of interest related to this
rationale behind the claim of microthread implants study.
maintaining crestal bone while translating higher stress
may be explained by the compressive nature of stress REFERENCES
translated by threads, as suggested by Oh et al.5 This 1. Adell R, Lekholm U, Rockler B, Brånemark PI. A 15-
remains to be verified in future in vitro studies. year study of osseointegrated implants in the treat-
ment of the edentulous jaw. Int J Oral Surg 1981;10:
In addition, our data suggested that when the abut- 387-416.
ment diameter was reduced by 10% or 20%, it resulted 2. Lekholm U, Adell R, Lindhe J, Brånemark PI. Marginal
in less stress transferred to the crestal bone, regard- tissue reactions at osseointegrated titanium fixtures
less of the type of thread pattern (microthread or (I). A 3-year longitudinal prospective study. Int J Oral
smooth) or direction of force (vertical or oblique). This Maxillofac Surg 1986;15:39-52.
3. Jemt T, Lekholm U, Gröndahl K. 3-year followup
is in support of the hypothesis that suggests that plat- study of early single implant restorations ad modum
form switching may increase the distance between the Brånemark. Int J Periodontics Restorative Dent 1990;
abutment inflammatory cell infiltrate and the alveolar 10:340-349.
crest, thus reducing the abutment design’s bone- 4. Buser D, Weber HP, Lang NP. Tissue integration of
resorptive effect.30 Furthermore, our results also non-submerged implants. 1-year results of a prospec-
tive study with 100 ITI hollow-cylinder and hollow-
confirmed the speculation made by Lazzara and screw implants. Clin Oral Implants Res 1990;1:33-40.
Porter,31 who found that repositioning the implant– 5. Oh TJ, Yoon J, Misch CE, Wang HL. The causes of
abutment interface away from the crestal bone into early implant bone loss: Myth or science? J Periodon-
a more confined area reduced bone resorption at tol 2002;73:322-333.
the crestal bone level. However, a recent animal 6. Albrektsson T, Zarb G. The long-term efficiency of
currently used dental implants: A review and proposed
study32 indicated that a smaller-diameter abutment criteria of success. Int J Oral Maxillofac Implants
(platform switching) reduced crestal bone loss only 1986;1:11-25.
for 28 days. After 28 days the matching and plat- 7. Misch CE. Dental Evaluation: Factors of Stress, 2nd ed.
form-switching abutments showed the same amount St. Louis: CV Mosby; 1999:122-123.
of bone loss. This finding was further supported by a 8. Jung YC, Han CH, Lee KW. A 1-year radiographic
evaluation of marginal bone around dental implants.
recent study published by Jung et al.,33 who showed Int J Oral Maxillofac Implants 1996;11:811-818.
implants with non-matching implant-abutment diam- 9. Guo E. Mechanical Properties of Cortical Bone and
eters (platform switching) demonstrated some bone Cancellous Bone Tissue, 2nd ed. Boca Raton, FL: CRC
loss; however, it was a small amount. The bone loss Press; 2001:1-23.
that occurred in the study was more related to implant 10. Rangert B, Krogh PH, Langer B, Van Roekel N. Bending
overload and implant fracture: A retrospective clinical
placement depth rather than to platform switching. analysis. Int J Oral Maxillofac Implants 1995;10:
For example, the greatest bone loss occurred when 326-334.
the implant-abutment junction was placed 1 mm be- 11. Natali AN, Pavan PG, Ruggero AL. Analysis of bone–
low the bone crest.33 These conflicting reports suggest implant interaction phenomena by using a numerical
that there is a need for more studies to validate the in- approach. Clin Oral Implants Res 2006;17:67-74.
12. Hayes WC. Biomechanics of Cortical and Trabecular
fluence of platform switching on crestal bone level. Bone: Implication for Assessment of Fracture Risk.
New York: Reven Press; 1991:130-144.
CONCLUSIONS 13. Pilliar RM, Desporter DA, Watson PA, Valiquette N.
Microthreads increase crestal stress upon loading. Dental implant design – Effect on bone remodeling.
When the concept of platform switching was applied J Biomed Mater Res 1991;25:467-483.
14. Vaillancourt H, Pilliar RM, McCammond D. Factors
by decreasing the abutment diameter, less stress affecting crestal bone loss with dental implants partially
was translated to the crestal bone in the microthread covered with a porous coating: A finite element anal-
and smooth-neck groups. Platform switching reduced ysis. Int J Oral Maxillofac Implants 1996;11:351-359.
stress to a greater degree in the microthread model 15. Ericsson I, Johansson CB, Bystedt H, Norton MR. A
compared to the smooth-neck model. histomorphometric evaluation of bone-to-implant
contact on machine-prepared and roughened titanium
dental implants. Clin Oral Implants Res 1994;5:202-
ACKNOWLEDGMENTS 206.
This study was partially supported by Student Re- 16. Abrahamsson I, Berglundh T. Tissue characteristics
search Program 2007-2008, School of Dentistry, Uni- at microthreaded implants: An experimental study in
versity of Michigan, and the Periodontal Graduate dogs. Clin Implant Dent Relat Res 2006;8:107-113.
17. Baumgarten H, Cocchetto R, Testori T, Meltzer A,
Student Research Fund. Project analysis was made Porter S. A new implant design for crestal bone
possible by Cranes Software. Vipul Kinariwala is the preservation: Initial observations and case report.
head of business development at Cranes Software, Pract Proced Aesthet Dent 2005;17:735-740.

2171
Effects of Microthreads and Platform Switching on Crestal Bone Stress Volume 79 • Number 11

18. Geng JP, Tan KB, Liu GR. Application of finite element 27. Hagberg C. Assessments of bite force: A review.
analysis in implant dentistry: A review of the literature. J Craniomandib Disord 1987;1:162-169.
J Prosthet Dent 2001;85:585-598. 28. Frost HM. A 2003 update of bone physiology and
19. O’Mahony A, Williams J, Spencer P. Anisotropic Wolff’s Law for clinicians. Angle Orthod 2004;74:3-15.
elasticity of cortical and cancellous bone in the pos- 29. Lee DW, Choi YS, Park KH, Kim CS, Moon IS. Effect of
terior mandible increases peri-implant stress and microthread on the maintenance of marginal bone
strain under oblique loading. Clin Oral Implants Res level: A 3-year prospective study. Clin Oral Implants
2001;12:648-657. Res 2007;18:465-470.
20. Sakaguchi RL, Borgersen SE. Nonlinear contact anal- 30. Ericsson I, Persson LG, Berglundh T, Marinello CP,
ysis of preload in dental implant screws. Int J Oral Lindhe J, Klinge B. Different types of inflammatory
Maxillofac Implants 1995;10:295-302. reactions in peri-implant soft tissues. J Clin Periodon-
21. Borchers L, Reichart P. Three-dimensional stress dis- tol 1995;22:255-261.
tribution around a dental implant at different stages of 31. Lazzara RJ, Porter SS. Platform switching: A new
interface development. J Dent Res 1983;62:155-159. concept in implant dentistry for controlling postrestor-
22. Rice JC, Cowin SC, Bowman JA. On the dependence ative crestal bone levels. Int J Periodontics Restorative
of the elasticity and strength of cancellous bone on Dent 2006;26:9-17.
apparent density. J Biomech 1988;21:155-166. 32. Becker J, Ferrari D, Herten M, Kirsch A, Schaer A,
23. Lekholm U, Zarb GA. Patient selection. In: Brånemark Schwaz F. Influence of platform switching on crestal
PI, Zarb GA, Albrektsson T, eds. Tissue-Integrated bone changes at non-submerged titanium implants: A
Prostheses: Osseointegration in Clinical Dentistry, 1st histomorphometrical study in dogs. J Clin Periodontol
ed. Chicago: Quintessence; 1985:117-128. 2007;34:1089-1096.
24. Kelly JR, Campbell SD, Bowen HK. Fracture-surface 33. Jung RE, Jones AA, Higginbottom FL, et al. The
analysis of dental ceramics. J Prosthet Dent 1989;62: influence of non-matching implant and abutment di-
536-541. ameters on radiographic crestal bone levels in dogs.
25. Holmgren EP, Seckinger RJ, Kilgren LM, Mante F. J Periodontol 2008;79:260-270.
Evaluating parameters of osseointegrated dental im-
plants using finite element analysis – A two-dimen- Correspondence: Dr. Hom-Lay Wang, Department of
sional comparative study examining the effects of Periodontics and Oral Medicine, School of Dentistry,
implant diameter, implant shape, and load direction. University of Michigan, 1101 N. University, Ann Arbor,
J Oral Implantol 1998;24:80-88. MI 48109-1078. Fax: 734/936-0374; e-mail: homlay@
26. Akagawa Y, Sato Y, Teixeira ER, Shindoi N, Wadamoto umich.edu.
M. A mimic osseointegrated implant model for three-
dimensional finite element analysis. J Oral Rehabil Submitted April 7, 2008; accepted for publication June 4,
2003;30:41-45. 2008.

2172