A statistical evaluation of microtensile bond strength

The purpose of this study was to determine the correlation between beams from the same tooth in a microtensile bond strength study and to examine their effect on the interpretation of re

a v a i l a b l e a t w w w s c i e n c e d i r e c t c o m

j o u r n a l h o m e p a g e : w w w i n t l e l s e v i e r h e a l t h c o m / j o u r n a l s / d e m a

A statistical evaluation of microtensile bond strength

methodology for dental adhesives

George J Eckerta, Jeffrey A Plattb, ∗

aDivision of Biostatistics, Indiana University School of Medicine, Indiana University Purdue University Indianapolis,

Indianapolis, IN, USA

bDepartment of Restorative Dentistry, Indiana University School of Dentistry, Indiana University Purdue University Indianapolis,

1121 West Michigan Street, Indianapolis, IN 46202, USA

a r t i c l e i n f o

Article history:

Received 14 October 2005

Received in revised form

18 January 2006

Accepted 3 February 2006

Keywords:

Microtensile bond strength

Adhesion

Adhesive incompatibility

Statistical methodology

a b s t r a c t

Objectives The purpose of this study was to determine the correlation between beams from

the same tooth in a microtensile bond strength study and to examine their effect on the interpretation of results

Methods A flat occlusal dentin surface was exposed using wet 180, 240, and 320-grit SiC

paper on caries-free human molar teeth Adhesive was applied followed by 6 mm of com-posite (CoreRestore2) applied in 2 mm increments Four adhesives were used: Optibond FL, light-cure, dual-cure, and self-cure Optibond Solo Plus Nine beams (1 mm2) were obtained from fifteen teeth for each group, labeled to indicate the orientation of the beams to each other and stored in artificial saliva at 37◦C for 1 week or 3 months Microtensile dentin bond strengths were determined using a non-trimming technique Statistical comparisons between adhesive-storage combinations were performed using ANOVA Analyses were per-formed assuming statistical independence between all beams and then repeated using a random tooth effect to account for correlations between beams from the same tooth

Results Significant correlations were found between beams No pattern was observed in

the correlations related to the orientations of the beams to each other Conclusions regard-ing statistical significance of comparisons (at alpha = 0.05) were sometimes affected by the incorrect assumption of independent beams The degree of correlation was large enough to severely affect power and sample size calculations

Significance Analyses of microtensile dentin bond strength studies need to account for

cor-relations between beams to avoid over-stating statistical significance of study results

© 2006 Academy of Dental Materials Published by Elsevier Ltd All rights reserved

1 Introduction

The evaluation of a dental adhesive/tooth interface often

involves an attempt to determine interfacial bond strength

Disagreement exists as to the most effective and beneficial

way to determine that bond strength The last decade has seen

a tremendous rise in the use of a methodology that has been

termed microtensile bond strength testing[1] The popularity

∗Corresponding author Tel.: +1 317 274 7229; fax: +1 317 278 7462.

E-mail address:jplatt2@iupui.edu(J.A Platt)

of the microtensile methodology is based on the idea that a better understanding of the strength of the adhesive interface can be obtained with smaller specimens The technique has been associated with fewer dentin cohesive failures and the ability to evaluate regional differences in bond strengths[2] This type of information may lead to more predictable restora-tive procedures and ultimately decrease the cost of dental care

0109-5641/$ – see front matter © 2006 Academy of Dental Materials Published by Elsevier Ltd All rights reserved

doi:10.1016/j.dental.2006.02.007

The methodology uses multiple specimens (∼1 mm2 in

cross-sectional area) that are obtained from one tooth either as

beams or dumbbell shaped specimens Comparisons between

adhesives or conditions are often made after pooling the

data from those specimens As many as 20 specimens per

tooth may be obtained[3]and the number of teeth used for

each group appear to have been as few as one[4] The data

from these investigations is often evaluated using a statistical

methodology known as analysis of variance (ANOVA)

Dur-ing 2004 and the first 6 months of 2005, at least 30 articles

were published using some form of parametric ANOVA[5–34]

This type of analysis assumes an independence of all of the

specimens included within each group being compared[35]

It is conceivable that, because the dentin substrate is not

uni-form, pooling of multiple data points from a single tooth is

acceptable methodology However, this assumption requires

a significant deviation from the accepted statistical method

and, as such, should be thoroughly validated Without that

validation, the results of many microtensile evaluations are

called into question

Dental adhesives have become more confusing to the

den-tal practitioner as more choices have become available An

example of this is a current adhesive that provides four

differ-ent choices; multi-compondiffer-ent, light-cured single compondiffer-ent,

dual-cured single component, and self-etch Advertisements

have stated that this provides the practitioner with “the

flex-ibility to choose your technique”[36] Recent concerns about

adhesive-core material compatibility[37]and adhesive

per-meability[38]may provide significant concerns when

select-ing the adhesive Significant portions of the data fuelselect-ing the

concerns have come from microtensile testing Furthermore,

the durability of some adhesives is adversely affected with

time[39] A discussion of the clinical implications of these

issues is reported elsewhere[40] In this study, the intent was

to provide the data necessary (1080 strength measurements)

to do an appropriate statistical analysis for evaluation of the

microtensile technique The various adhesives were selected

with the expectation that differences between mean values

would be seen The hypothesis tested was that there is a

correlation between beams taken from the same tooth for

microtensile testing that will impact the statistical

interpre-tation of the results

2 Materials and methods

Eight groups of 15 teeth were fabricated for this study (Table 1)

Human molar teeth collected under an IUPUI/Clarion IRB

approved protocol, were treated in 10% formalin for <2 weeks and stored in deionized water until use Caries free, non-restored teeth were selected, hand scaled and refrigerated

A flat occlusal dentin surface was exposed using 180-grit SiC paper under de-ionized water flow The absence of enamel was verified using a stereomicroscope

2.1 Adhesives and composite

All restorative materials for all groups were from one com-pany, sds Kerr, Orange, CA (Table 2) The dual-cured resin-matrix composite core material for all groups was CoreRe-store2 Base Universal and CoreReCoreRe-store2 Catalyst High mixed

in equal amounts Polymerization was accomplished using an ESPE Highlight QTH light-activating unit (800 mW/cm2)

2.2 Groups 1 and 5

The dentin was etched with a 37% phosphoric acid gel for 15 s, rinsed for 15 s and blot dried Optibond FL primer was placed for 15 s and air dried for 3 s Optibond FL adhesive was applied and light activated for 30 s

2.3 Groups 2 and 6

The dentin was etched with a 37% phosphoric acid gel for 15 s, rinsed for 15 s and blot dried Optibond Solo Plus was applied with a microbrush for 15 s with a light brushing motion and air-thinned for 3 s with canned compressed air to achieve a visibly uniform layer The adhesive was light-cured for 20 s

2.4 Groups 3 and 7

The dentin was etched with a 37% phosphoric acid gel for 15 s, rinsed for 15 s and blot dried One drop of Optibond Solo Plus and one drop of Optibond Solo Plus Activator were placed in

a mixing well, mixed for 3 s, applied with a microbrush for

15 s with a light brushing motion and air-thinned for 3 s with canned compressed air to achieve a visibly uniform layer The adhesive was light-cured for 20 s

2.5 Groups 4 and 8

Self-Etch Primer was applied to the dentin for 15 s with a light brushing motion and air-thinned for 3 s Optibond Solo Plus was then applied for 15 s with a light brushing motion and air-thinned for 3 s to achieve a visibly uniform layer The adhesive was light-cured for 20 s

Table 1 – Eight groups prepared for evaluation by the indicated test methodology

Table 2 – Adhesives used in the study

Optibond FL (Kerr, Orange, CA, USA)

water, PAMA

Etch for 15 s with 37.5% phosphoric acid (Kerr gel Etchant) Rinse for 15 s Blot dry Apply prime and rub for 15 s Dry for 5 s Apply adhesive in a uniform thin layer Light cure for 30 s

ODMAB, S530, A174, OX-50, SP345,

Na2Si6F Optibond Solo Plus (Kerr)

Prime/adhesive Bis-GMA, HEMA, GDM, GPDM,

EtOH, CQ, ODMAB, BHT, TS530, A174, SP345, Na2Si6F

Etch for 15 s with Kerr gel Etchant Rinse for

15 s Blot dry Apply prime/adhesive and rub for 15 s to achieve a uniform layer Gently air blow for 3 s and light cure for 20 s Optibond Solo Plus dual cure (Kerr)

Prime/adhesive Same as Optibond Solo Plus Etch, rinse and dry as above Mix one drop

of Optibond Solo Plus prime/adhesive and one drop of activator for 3 s Apply the mixture and rub for 15 s to achieve a uniform layer Light cure for 20 s

Optibond Solo Plus self-etch (Kerr)

Self-etch primer HFGA-GDM, GPDM, EtOH, MEHQ,

ODMAB, CQ

Apply self-etch primer and rub for 15 s Gently air blow for 3 s Apply Optibond Solo Plus and rub for 15 s to achieve a uniform layer Light cure for 20 s

Information given from the manufacturer (Kerr); A174 = gamma-methacryloxypropyltrimethoxysilane; bis-GMA =

bis-phenol-A-bis-(2-hydroxy-3-methacryloxypropyl)ether; BHT = 2,6-di-(tert-butyl)-4-methylphenol; BSA = benzene sulfinic acid sodium salt; CQ =

1,7,7-trimethylbicyclo-[2,2,1]-hepta-2,3-dione; EtOH = ethanol; GDM = glycerol dimethacrylate; GPDM = glycerol phosphate dimethacrylate; HEMA = 2-hydroxyethylmethacrylate; HFGA-GDM = hexafluoroglutaric anhydride-Glyceroldimethacrylate adduct; MEHQ = 4-methoxyphenol;

Na2Si6F = disodium hexafluorosilicate; ODMAB = 2-(ethylhexyl)-4-(dimetylamino)benzoate; OX-50 = fumed silicon dioxide; PAMA = phthailic acid monomethacrylate; SP345 = barium aluminoborosilicate; TS530 = fumed silicon dioxide

Just prior to specimen fabrication, the dentin surfaces were

prepared to 320-grit and subjected to ultrasonic cleaning for

10 min The adhesives were applied and a mass of composite

6 mm high created by light-activating 2 mm increments, each

for 40 s Twenty-four hours after fabrication, each of the teeth

were sectioned in two planes, 90◦to each other, to obtain nine

beams with a cross-sectional area <1 mm2 Thus, each group

was planned for 135 beams Each beam was labeled so that

knowledge of the orientation of the beams to each other was

maintained

Specimens were stored in artificial saliva at 37◦C until

test-ing One group for each adhesive was tested at 1 week and the

other was tested after three months of storage The artificial

saliva for the three-month group was changed during each

week of storage to minimize the risk of bacterial growth

Test-ing was accomplished usTest-ing a MTS Sintech Renew Universal

Testing machine operating with TestWorks 4 Each beam was

attached to a modified Bencor Multi-T testing device using a

cyanoacrylate adhesive Testing occurred at a crosshead speed

of 1 mm/min

2.6 Statistical methods

Comparisons between the four adhesives and the two

stor-age periods for differences in microtensile bond strength were

performed using repeated measures ANOVA Factors for

adhe-sive type, storage period, and the adheadhe-sive-by-storage interac-tion were included in the ANOVA model, as well as a random effect to allow correlation between the nine beams from each tooth Several variance/covariance structures were examined

to determine if the beams correlated in a general fashion

or correlated in a manner dependent upon proximity of the beams The analyses were repeated under the assumption

of independence between beams to examine how much the results were affected by ignoring the correlations Fisher’s pro-tected least significant differences was used to control the overall significance level for the comparisons at 5% Many beams were unable to be tested due to debonding before placement on the testing machine Analyses were performed excluding these beams and again including these beams with

a microtensile bond strength of zero

Statistical analysis of bond strength measurements are sometimes performed using survival analysis rather than ANOVA, using force required for bond failure in place of the usual ‘time to event’ seen in typical survival analyses[41,42] The survival analysis can be performed using a parametric Weibull model or a non-parametric Cox proportional hazards model Survival analyses can accommodate a random effect

to accommodate the within-tooth correlations, and such sur-vival analysis models are often called frailty models[43,44] It

is not uncommon for some slight differences in the compar-isons when analyzing the bond strengths using ANOVA and

Table 3 – Correlations between beams

4 0.18 0.29 0.14

5 0.17 0.25 0.39 0.25

6 0.23 0.45 0.46 0.38 0.49

7 0.12 0.18 0.33 0.22 0.17 0.27

8 0.20 0.22 0.33 0.38 0.50 0.45 0.17

9 0.00 0.13 0.04 0.00 0.12 0.19 0.51 0.22

survival analysis because of the different assumptions used

in the two methods, even when both methods appear to be

appropriate

2.7 Sample size justification

Micro-tensile bond strength: from the study by Tay et al.[38],

we estimated that the within-group standard deviation would

be 3.7 Because the actual dependence between the beams was

unknown, a range of detectable differences was used,

assum-ing total dependence and total independence between the

beams With a sample size of 15 teeth per group, the study

was predicted to have 80% power to detect a difference of

1.3–4.0 between groups If the adhesive-by-storage interaction

was not significant, the study would have been able to detect

differences of 0.9–2.8 between adhesives and differences of

0.8–2.3 between storage periods The total sample size of 90

teeth for the micro-tensile bond strength analyses was

suf-ficient to provide good estimates of the variance/covariance

matrix used in determining the independence of the beams

A 99% lower confidence bound for the correlation coefficient

did not include zero for correlations of at least 0.21

3 Results

Results presented here are from analyses using only beams

which did not debond before microtensile testing The groups

with higher proportions of spontaneously debonded beams

also had lower microtensile bond strengths for the beams,

which did not spontaneously debond, so similar results

were obtained when spontaneously debonded beams were

included as zero bond strength

The correlation matrix from the repeated measures ANOVA

with an unstructured variance/covariance/correlation matrix

(Table 3) shows the correlations between beams from the same

tooth There does not appear to be a general pattern to the

cor-relations, so the analyses presented assume a general

corre-lation between beams The overall correcorre-lation between beams

was estimated to be 0.27

Microtensile bond strength (Table 4 and Fig 1) was

sig-nificantly higher for 1WOP than 3MOP Microtensile bond

strength was marginally higher for 3MOF than for 1WOF when

compared using ANOVA but not different when using

sur-vival analysis There was no significant storage time effect

on microtensile bond strength seen with 1WOD and 3MOD or

with 1WOS and 3MOS (Fig 2,Table 5)

Microtensile bond strength was significantly lower for

1WOD than for 1WOF, 1WOP, and 1WOS Also, microtensile

Table 4 – Microtensile bond strength summary statistics

by group

1WOF 111 45.2 2.3 1.6 12.1 96.3 44.8

SE1: standard error from ANOVA with a compound-symmetry variance/covariance (general correlation) matrix; SE2: standard error assuming beams independent

Fig 1 – Microtensile bond strength: mean + 1 standard error.

bond strength was significantly lower for 1WOS than for 1WOF and 1WOP Microtensile bond strength was significantly lower for 1WOP than for 1WOF when compared using ANOVA but not different when using survival analysis

Microtensile bond strength was significantly lower for 3MOD than for 3MOF and 3MOP; in the survival

analy-Fig 2 – Microtensile bond strength: estimated survival functions.

Table 5 – Microtensile bond strength p-values for comparisons assuming correlation or independence between beams

from the same tooth

Storage time comparisons

Group comparisons

sis 3MOD was also lower than 3MOS Also, microtensile

bond strength was significantly lower for 3MOS than for

3MOF and 3MOP, and 3MOP was significantly lower than

3MOF

3.1 Effect of assuming beam independence

The analyses assuming independence between beams

over-state statistical significance levels for a number of

com-parisons, several of which would lead to incorrectly

deter-mining comparisons to be significantly different when they

should not be considered significantly different (Table 5) If

incorrectly assuming beams obtained from the same tooth

were statistically independent, and assuming ANOVA-based

analyses with a 5% significance level and 80% power, to

detect a difference of 7.5 between two groups would require

a sample size of 65 beams per group (or eight teeth per

group sectioned into nine beams per tooth) When

prop-erly accounting for the within-tooth correlation, the number

of specimens required to detect a difference of 7.5 with at

least 80% power is 23 teeth (207 beams) per group [45] Or

looked at differently, the power to detect a difference of 7.5

is reduced to 33% when using eight teeth (72 beams) per

group

4 Discussion

Although the correlation between beams near 0.30 would not

be considered a ‘strong’ correlation, the effect of this level

of correlation is magnified by the number of beams obtained

from each tooth Multiple beams from a tooth coupled with

even this weak to slight correlation will affect the

signifi-cance of the group comparisons In certain cases, the corre-lations would improve rather than reduce the power of the study if the study design is such that beams from the same teeth can be placed into different treatment groups Whether

or not the conclusions would change for any specific study because of the within-tooth correlation would be unknown before the study was performed It would depend on the num-bers of beams per tooth and teeth per group (i.e sample size) and on the actual group means observed in the study The hypothesis for this investigation was supported by the results

The results presented in this paper use a simple random effect for tooth in the ANOVA to account for the correla-tions Other options are available using repeated measures ANOVA that would allow beam-specific correlations (Table 3)

or a ‘spatial’ correlation matrix that uses the distance between the beams to determine the correlation While the general correlation was sufficient for our study, other studies may require more complex correlations between beams because

of differences in beam orientation, beam shape, or materials used

4.1 Significance

Analyses of microtensile dentin bond strength studies need to account for correlations between beams to avoid over-stating statistical significance of study results

Acknowledgement

This work was supported in part by a grant from Delta Dental Fund

r e f e r e n c e s

[1] Sano H, Shono T, Sonoda H, Takatsu T, Ciucchi B, Carvalho R,

et al Relationship between surface area for adhesion and

tensile bond strength—evaluation of a micro tensile bond

test Dent Mater 1994;10:236–40

[2] Shono Y, Ogawa T, Terashita M, Carvalho RM, Pashley EL,

Pashley DH Regional measurement of resin–dentin bonding

as an array J Dent Res 1999;78:699–

705

[3] Papacchini F, Goracci C, Sadek FT, Monticelli F, Garcia-Godoy

F, Ferrari M Microtensile bond strength to ground enamel by

glass-ionomers, resin-modified glass ionomers, and resin

composites used as pit and fissure sealants J Dent

2005;33:459–67

[4] Yamada M, Miyazaki M, Moore BK Influence of

interchanging adhesive resins and self-etching primers on

the mechanical properties of adhesive resins Oper Dent

2004;29:532–7

[5] Giannini M, Soares CH, de Carvalho RM Ultimate tensile

strength of tooth structures Dent Mater 2004;20:

322–9

[6] Filho AM, Vieira LC, Araujo E, Monteiro Jr S Effect of

different ceramic surface treatments on resin microtensile

bond strength J Prosthodont 2004;13:28–35

[7] Bedran-de-Castro AK, Pereira PN, Pimenta LA, Thompson JY

Effect of thermal and mechanical load cycling on

microtensile bond strength of a total-etch adhesive system

Oper Dent 2004;29:150–6

[8] Yazici AR, Akca T, Ozgunaltay G, Dayangac B Bond strength

of a self-etching adhesive system to caries-affected dentin

Oper Dent 2004;29:176–81

[9] El Zohairy AA, de Gee AJ, de Jager N, van Ruijven LJ, Feilzer

AJ The influence of specimen attachment and dimension

on microtensile strength J Dent Res 2004;83:

420–4

[10] Reis A, Rocha de Oliveira Carrilho M, Schroeder M, Tancredo

LL, Loguercio AD The influence of storage time and cutting

speed on microtensile bond strength J Adhes Dent

2004;6:7–11

[11] Carvalho RM, Tay FR, Giannini M, Pashley DH Effects of

pre-and post-bonding hydration on bond strength to dentin J

Adhes Dent 2004;6:13–7

[12] Lopes GC, Baratieri CM, Baratieri LN, Monteiro Jr S, Cardoso

Vieira LC Bonding to cervical sclerotic dentin: effect of acid

etching time J Adhes Dent 2004;6:

19–23

[13] Osorio R, Toledano M, Osorio E, Tay FR Longevity of bonds

made by composite and polyacid-modified resins to dentin

using a dual-cured adhesive system Am J Dent

2005;18:19–22

[14] Doi J, Itota T, Yoshiyama M, Tay FR, Pashley DH Bonding to

root caries by a self-etching adhesive system containing

MDPB Am J Dent 2004;17:89–93

[15] Kelsey III WP, Latta MA, Vargas MA, Carroll LR, Armstrong

SR Microtensile bond strength of total-etch and self-etch

adhesives to the enamel walls of Class V cavities Am J Dent

2005;18:37–40

[16] Tay FR, Pashley DH, King NM, Carvalho RM, Tsai J, Lai SC, et

al Aggresiveness of self-etch adhesives on unground

enamel Oper Dent 2004;29:309–16

[17] Erdemir A, Eldeniz AU, Belli S, Pashley DH Effect of solvents

on bonding to root canal dentin J Endod 2004:589–92

[18] Choi KK, Ryu GJ, Choi SM, Lee MJ, Park SJ, Ferracane JL

Effects of cavity configuration on composite restoration

Oper Dent 2004;29:462–9

[19] Koase K, Inoue S, Noda M, Tanaka T, Kawamoto C, Takahashi

A, et al Effect of bur-cut dentin on bond strength using two

all-in-one and one two-step adhesive systems J Adhes Dent 2004;6:97–104

[20] Reis AF, Giannini M, Kavaguchi A, Soares CJ, Line SR Comparison of microtensile bond strength to enamel and dentin of human, bovine, and procine teeth J Adhes Dent 2004;6:121–77

[21] Lopes GC, Cardoso Pde C, Vieira LC, Baratieri LN Microtensile bond strength to root canal vs pulp chamber dentin: effect

of bonding strategies J Adhes Dent 2004;6:129–33

[22] Cavalli V, Giannini M, Carvalho RM Effect of carbamide peroxide bleaching agents on tensile strength of human enamel Dent Mater 2004;20:733–9

[23] Yamada M, Miyazaki M, Moore BK Influence of interchanging adhesive resins and self-etching primers on the mechanical properties of adhesive resins Oper Dent 2004;29:532–7

[24] Dias WR, Pereira PN, Swift Jr EJ Effect of bur type on microtensile bond strengths of self-etching systems to human dentin J Adhes Dent 2004;6:195–203

[25] Trajtenberg CP, Pereira PN, Powers JM Resin bond strength and micromorphology of human teeth prepared with an Erbium:YAG laser Am J Dent 2004;17:331–6

[26] Kanno T, Ogata M, Foxton RM, Nakajima M, Tagami J, Miura

H Microtensile bond strength of dual-cure resin cement to root canal dentin with different curing strategies Dent Mater J 2004;23:550–6

[27] Loguercio AD, Uceda-Gomez N, de Oliveira Carrilho MR, Reis

A Influence of specimen size and regional variation on long-term resin-dentin bond strength Dent Mater 2005:224–31

[28] Bottino MA, Valandro LF, Scotti R, Buso L Effect of surface treatments on the resin bond to zirconium-based ceramic Int J Prosthodont 2005;18:60–5

[29] De Munck J, Vargas M, Iracki J, Van Landuyt K, Poitevin A, Lambrechts P, et al., Van Meerbeek B One-day bonding effectiveness of new self-etch adhesives to bur-cut enamel and dentin Oper Dent 2005;30:39–49

[30] Coelho Santos MJ, Navarro MF, Tam L, McComb D The effect

of dentin adhesive and cure mode on film thickness and microtensile bond strength to dentin in indirect restorations Oper Dent 2005;30:50–7

[31] Say EC, Nakajima M, Senawongse P, Soyman M, Ozer F, Tagami J Bonding to sound vs caries-affected dentin using photo- and dual-cure adhesives Oper Dent 2005;30:

90–8

[32] Reis A, de Oliveira Bauer JR, Loguercio AD Influence of cresshead speed on resin-dentin microtensile bond strength J Adhes Dent 2004;6:275–8

[33] Goracci C, Sadek FT, Monticelli F, Cardoso PE, Ferrari M Microtensile bond strength of self-etching adhesives to enamel and dentin J Adhes Dent 2004;6:313–8

[34] Valandro LF, Leite FP, Scotti R, Bottino MA, Niesser MP Effect

of ceramic surface treatment on the microtensile bond strength between a resin cement and an alumina-based ceramic J Adhes Dent 2004;6:327–32

[35] Dawson B, Trapp RG Basic and clinical biostatistics New York: Lange Medical Books/McGraw Hill; 2004

p 170

[36] http://www.kerrdental.com/OptibondSoloPlusSelfEtch/index htmlApril 10, 2003

[37] Sanares AM, Itthagarun A, King NM, Tay FR, Pashley DH Adverse surface interactions between one-bottle light-cured adhesives and chemical-cured composites Dent Mater 2001;17:542–56

[38] Tay FR, Pashley DH, Suh BI, Carvalho RM, Itthagarun A Single-step adhesives are permeable membranes J Dent 2002;30:371–82

[39] Giannini M, Seixas CA, Reis AF, Pimenta LA Six-month storage-time evaluation of one-bottle adhesive systems to

dentin J Esthet Restor Dent 2003;15:43–8, discussion

49

[40] Chang J, Platt JA, Yi K, Cochran MA Quantitative comparison

of the water permeable zone among four types of dental

adhesives used with a dual-cured composite Oper Dent

2006;31(3); in press

[41] Armstrong SR, Vargas MA, Fang Q, Laffoon JE Microtensile

bond strength of a total-etch 3-step, total-etch 2-step,

self-etch 2-step, and a self-etch 1-step dentin bonding

system through 15-month water storage J Adhes Dent

2003;5(1):47–56

[42] Pilecki P, Stone DG, Sherriff M, Watson TF Microtensile bond strengths to enamel of self-etching and one bottle adhesive systems J Oral Rehab 2005;32(7):531–40

[43] Therneau TM, Grambsch PM Modeling survival data: extending the Cox model New York: Springer-Verlag; 2000

[44] Klein JP, Moeschberger ML Survival analysis: techniques for censored and truncated data New York: Springer-Verlag; 2003

[45] Liu G, Liang KY Sample size calculations for studies with correlated observations Biometrics 1997;53:937–47