Clinical Research

Treatment Outcome in Endodontics: The Toronto Study.

Phase III: Initial Treatment
Vincent L. Marquis, DMD, MSc, Thuan Dao, DMD, MSc, PhD, Mahsa Farzaneh, DDS, MSc,
Sarah Abitbol, DDS, MSc, and Shimon Friedman, DMD

Abstract
The 4- to 6-year outcome of initial endodontic treat-
ment was assessed for phase III (1998-1999) of the
Toronto Study. Of the 532 teeth treated, 248 were from
E ndodontic treatment is performed to prevent or cure apical periodontitis and to
retain the treated tooth in function. The predictability of achieving these goals has
recently been questioned (1–3), as the inconsistent outcomes reported for endodontic
discontinuers (excluded), 142 from dropouts, 10 ex- treatment contrast with consistently favorable outcomes reported for implant-sup-
tracted, and 132 (50% recall) examined for outcome: ported single-tooth replacement. Thus, the outcome of endodontic treatment has re-
healed (no apical periodontitis [AP], signs, symptoms) cently come to the forefront as the focus of a debate regarding tooth retention or
or diseased. Phase III was analyzed alone and com- replacement.
bined with phases I, II (n ⫽ 373 teeth). Logistic regres- The outcome of initial (first-time) endodontic treatment has been assessed in
sion performed on the combined phases I-III sample many studies during the past decades; however, the results have varied considerably
identified significant (p ⱕ 0.05) outcome predictors: (4). This wide variation of the reported outcomes, attributed mainly to differences in
preoperative AP (OR ⫽ 3.5; CI 1.7-7.2; healed: absent, methodology (4), has caused considerable confusion in the profession, and it interferes
93%; present, 80%), number of roots (OR ⫽ 2.2; CI with attempts to establish evidence-based guidelines for endodontic practice. To pro-
1.0-4.7; healed: 1 - 92%; ⱖ2 - 83%), and intraopera- vide the evidence base to support endodontic treatment, outcome studies must conform
tive complications (OR ⫽ 2.2; CI 1.1-4.5; healed: ab- to design and methodology criteria consistent with an adequate (at least mid-range)
sent, 88%; present, 76%). Treatment technique (OR ⫽ level of evidence (5, 6), e.g. they should be randomized controlled trials or method-
2.8; CI 1.3-6.1; healed: Schilder, 89%; alternative, 73%) ologically sound observational cohort studies. A recent review (7) has highlighted the
was suggested as an outcome predictor in teeth with shortage of such studies, identifying only 14 (8 –21) that appear to methodologically
AP, requiring confirmation from randomized controlled conform to the mid-range level of evidence. Adding three more recent studies (22–24),
trials. (J Endod 2006;32:299 –306) the short list now comprises 17 studies. However, because several of these studies
(8 –12) include treatment techniques that have been modified in the recent years, they
Key Words may no longer be considered as reflecting the outcome of state-of-the-art endodontic
Apical periodontitis, endodontic treatment, prognosis, treatment (4, 7). The remaining 12 studies comprise the evidence base for current
root canal therapy, treatment outcome
initial endodontic treatment; clearly, additional studies are required to broaden this
evidence base.
Of the current studies, two have reported on phases I (20) and II (21) of the
From the MSc Program in Endodontics, University of To- prospective Toronto Study project. This modular project was initiated in 1993 and
ronto, Toronto, Ontario, Canada. designed as a continuous investigation of the 4- to 6-year outcome of endodontic
Address requests for reprints to Dr. Shimon Friedman, treatment performed by graduate endodontics students. Patients have been recalled in
Department of Graduate Endodontics, University of Toronto, 2-year “phases,” and the sample size multiplied with each successive phase added, so
124 Edward Street, Room 348, Toronto, Ontario, Canada.
0099-2399/$0 - see front matter
as to increase the statistical power for assessment of significant outcome predictors.
Copyright © 2006 by the American Association of Thus, the analyzed sample for initial treatment has multiplied from 120 teeth in phase
Endodontists. I (20) to 242 teeth in phases I and II combined (21). Consequently, the number of
doi:10.1016/j.joen.2005.10.050 significant outcome predictors suggested by a multivariate analysis increased from one
(20) to two (21). Preoperative apical periodontitis (AP), shown by the majority of
previous studies (8 –10, 12–14, 18) to influence the outcome of treatment, was clearly
identified as a significant outcome predictor. In addition, treatment technique was
highlighted as warranting investigation in randomized controlled trials, with a better
outcome observed after flared canal preparation and vertical compaction of warm
gutta-percha, as described by Schilder (25, 26), than after step-back canal preparation
and lateral compaction of gutta-percha. It was expected that by extending the Toronto
Study project onto an additional phase, the increased sample size and statistical power
would allow corroboration of the previously identified outcome predictors and the
identification of additional ones.
The objectives of this study were 2-fold: (a) to systematically assess the 4- to 6-year
outcome of initial endodontic treatment in phase III of the Toronto Study project, and
(b) to assess associations between the outcome of treatment and pre-, intra-, and
postoperative variables, by combining the phase III sample with those of the previous
two phases for increased statistical power.

JOE — Volume 32, Number 4, April 2006 Treatment Outcome in Endodontics 299
Clinical Research
Materials and Methods Analysis
The protocol of the Toronto Study project was established before Separate statistical analyses were performed on the data of this
subjects were recruited and treated. The study protocol and informed study (phase III) and on the pooled data from phases I, II, and III.
consent forms were approved by the University of Toronto Health Sci- Pooling of the data was justified by the consistency of the methodology
ences Research Ethics Board. The same methodology as described in for all phases of the Toronto Study project. Univariate analysis (percent
the previous reports (20, 21) was followed in this study. It is briefly frequencies) characterized the data. Bivariate analysis (␹2 test of pro-
summarized below. portions and Fisher’s exact test) examined associations between the
treatment outcome and pre-, intra-, and postoperative variables, to sug-
1. Cohort: The inception cohort comprised all 532 teeth in 468 gest potentially important variables for inclusion in the multivariate
patients who had received initial endodontic treatment from Jan- analysis. Multivariate analysis (logistic regression) was performed on
uary 1998 to December 1999. the pooled data only, incorporating the variables found significant in the
2. Intervention: Supervised graduate students provided treatment in bivariate analysis into a prediction model, in order to identify significant
accordance with two main treatment techniques: (a) step-back/ outcome predictors. The dependent variable in all analyses was the
lateral compaction (SBLC), and (b) flared preparation/vertical dichotomous outcome, healed versus disease. All tests were performed
compaction (FPVC). Ten teeth in total were treated with different as two-tailed and interpreted at the 5% significance level. In addition to
the analysis of the complete sample, stratified analyses were performed
root filling techniques, either single gutta-percha cone with a
for teeth treated without or with preoperative radiolucency. Table 1 lists
glass-ionomer cement sealer (Ketac-Endo, 3M ESPE, St. Paul,
all of the 17 investigated variables.
MN), or injectable gutta-percha (Obtura II, Obtura Spartan, Fen-
ton, MO). Each technique was performed only in specific ses-
sions during the week. Allocation of treatment techniques was Results
quasirandomized by allowing patients to select clinical sessions Unweighted Cohen’s kappa scores for the intraexaminer agreement of
according to their availability and convenience. the phase III examiner (␬ ⫽ 0.96) indicated “very good agreement” (28).
The intervention in the phase III cohort differed from that in The interexaminer agreement between the phase III examiner and the phase
phases I and II in three ways: (a) engine-driven NiTi instruments I and II examiners (␬ ⫽ 0.67 and 0.63, respectively), and the co-investi-
were used routinely rather than hand stainless steel files; (b) The gator (␬ ⫽ 0.65) indicated “good agreement” (28).
FPVC technique was modified from the classic Schilder technique
(25, 26) by treating all canals concurrently rather than in se- Phase III Sample
quence, by using the System B (SybronEndo, Orange, CA) and The inception cohort of 468 patients and 532 teeth was distributed into
Obtura II (Obtura Spartan) devices rather than sectional gutta- the following categories: (a) discontinuers (excluded)—248 teeth from
percha and conventional heat carriers, and by using engine- 222 relocated subjects who could not be contacted; (b) dropouts—142
driven instruments as mentioned above; and (c) operating mi- teeth from 21 subjects who declined the recall and 105 subjects who did not
croscopes were used routinely in all treated cases, rather than respond; and (c) attending—142 teeth (50% recall rate after exclusion of
loupes. discontinuers) from 120 subjects, including 132 teeth examined for out-
All preoperative and intraoperative data were uniformly recorded come (study sample) and 10 extracted teeth (five for restorative consider-
by the providers of treatment and entered into a database in real ations, three for fractures, one for advanced periodontal disease, and one
time. for unknown reasons). The examined study sample is compared to the
3. Recall: All subjects were recalled by letters, invited to attend a inception cohort in Table 1. Response bias analysis (not shown) suggested
follow-up examination, and offered compensation for work time that the attending and lost-to-follow-up (dropouts and discontinuers) pop-
lost and travel expenses incurred by attending. Nonresponders ulations differed significantly with regards to age.
were contacted by telephone and encouraged to attend. When At the end point of the phase III study, 113/132 teeth (86%) were
treated teeth were reported to be lost, subjects were questioned classified as healed. One of these teeth presented with vertical root fracture
and the records of those who received regular care at the Faculty associated with severe bone loss, and was excluded from further analysis.
of Dentistry examined, to establish the reasons for extraction. Nineteen teeth (14%) were classified as having disease. The bivariate anal-
4. Outcome assessment: All follow-up examinations were per- ysis (Table 2) suggested only one statistically significant association, with a
formed by the phase III-designated examiner (V.M.). Before ex- higher healed rate for teeth without than with preoperative radiolucency.
amining subjects, he was calibrated for use of the Periapical Four additional variables (tooth type, number of treatment sessions, root
Index (PAI) (27) in the same manner as the examiners for phases filling voids, and type of coronal seal material) were associated with healed
I and II and the co-investigator (S.F.) for the project. Interexam- rate differentials of ⱖ10%, which were not statistically significant.
iner and intraexaminer agreement scores were calculated by us- Of the 112 teeth classified as healed (one fractured tooth excluded),
ing weighted Cohen’s kappa statistics. PAI scores were dichoto- seven teeth (6%) presented with a slight tenderness to percussion. Of the 19
mized to reflect absence (PAI ⱕ 2) or presence (PAI ⱖ 3) of AP. teeth classified as having disease, three teeth (16%) presented signs and
The evaluated unit was the tooth as a whole, with multirooted symptoms (pain), of which two had a PAI score ⱕ2 (no AP). Thus in total,
teeth assigned the highest score of all roots. All postoperative data 121 teeth (105 healed and 16 having disease) of 131 analyzed (92%) were
recorded at the follow-up examination were immediately entered fully functional, without any signs or symptoms and without tenderness to
into the database. percussion. In the 16 teeth with PAI ⱖ 3, the lesion was smaller than
preoperatively in four teeth (25%), unchanged in five teeth (31%), and
Based on clinical and radiographic measures, the outcome was increased or new in seven teeth (44%).
dichotomized either as “healed” (absence of AP, signs and symptoms Sixty-one of 65 teeth (94%) without preoperative radiolucency
other than tenderness to percussion), or as “disease” (presence of healed. Stratified analysis showed no variables associated with sta-
either AP, signs or symptoms). Teeth presenting without clinical signs or tistically significant or large healed rate differentials. Of 66 teeth
symptoms were considered “functional” regardless of the PAI score. with preoperative radiolucency, 51 teeth (77%) healed. Stratified

300 Marquis et al. JOE — Volume 32, Number 4, April 2006

 Clinical Research
TABLE 1. Univariate distribution of investigated factors in the study population
Phase III Phases I-III Pooled
Prognostic Factor Inception Cohort Study Sample Inception Cohort Study Sample
% (n ⴝ 532) % (n ⴝ 132) % (n ⴝ 1370) % (n ⴝ 373)
Preoperative
Age
ⱕ45 yr 47 21 50 25
⬎45 yr 53 77 50 75
Gender
Female 58 59 55 55
Male 42 41 45 45
Tooth type
Anterior 16 16 18 18
Posterior 84 84 82 82
Tooth location
Maxilla 50 46 51 50
Mandible 50 54 49 50
No. of roots
1 27 30 29 31
ⱖ2 73 70 71 69
Signs and symptoms
Absent 41 45 37 38
Present 59 55 63 62
Radiolucency
Absent 46 50 43 43
Present 54 50 57 57
Pulp status
Responsive 33 42 34 36
Nonresponsive 67 58 66 64
Intraoperative
Treatment sessions
1 15 17 18 18
ⱖ2 85 83 82 82
Technique
SBLC 54 55 52 51
FPVC 43 43 45 45
Other* 3 2 3 4
Root-filling length
Adequate 74 73 76 74
Short 8 10 7 7
Long 18 17 17 19
Root-filling voids
Absent 90 86 85 84
Present 10 14 15 16
Sealer extrusion
Absent 38 42 45 46
Present 62 58 55 54
Complications
Absent 83 79 84 83
Present 17 21 16 17
Coronal seal material
Temporary† 12 15 15 18
Definitive‡ 82 85 85 82
Postoperative
Restoration at follow-up
Temporary 2 5
Definitive filling 25 28
Crown 73 67
Post
Absent 70 58
Present 30 42
SBLC ⫽ Modified step-back preparation, lateral compaction of gutta-percha; FPVC ⫽ Flared preparation, vertical compaction of warm gutta-percha.
*Modified step-back preparation, single gutta-percha cone and Ketac-Endo sealer.
†Cavit, ZOE, IRM.
‡Amalgam, composite resin, glass ionomer cement, crown.

analysis (Table 3) suggested only two statistically significant asso- defect, number of treatment sessions, type of sealer, sealer extru-
ciations, with a higher healed rate for maxillary than mandibular sion, complications, and restoration at follow-up) were associated
teeth, and for coronal seal with a definitive than a temporary restor- with healed rate differentials of ⱖ10% that were not statistically
ative material. Seven additional variables (tooth type, periodontal significant.

JOE — Volume 32, Number 4, April 2006 Treatment Outcome in Endodontics 301
Clinical Research
TABLE 2. Bivariate analysis of associations between selected factors§ and the healed rate 4 to 6 yr after treatment, presented for phase III (n ⫽ 131) and the
pooled phases I–III (n ⫽ 369)
Phase III Phase I-III Combined
Prognostic Factor Healed Healed
n p Value n p Value
(% n) (% n)
Preoperative
Gender
Female 201 89 0.041
Male 168 82
Tooth type
Anterior 21 95 0.167
Posterior 110 84
No. of roots
1 114 92 0.018
ⱖ2 255 83
Radiolucency
Absent 65 94 0.007 160 93 < 0.001
Present 66 77 209 80
Intraoperative
Treatment sessions
1 22 96 0.195*
ⱖ2 109 84
Root-filling voids
Absent 113 88 0.140*
Present 18 72
Complications
Absent 306 88 0.019
Present 63 76
Coronal seal material
Temporary† 20 75 0.169*
Definitive‡ 111 87
Bold type face highlights statistical significance.
*Fischer’s Exact test, ␹2 test otherwise.
†Cavit, ZOE, IRM.
‡Amalgam, composite resin, glass ionomer cement, crown.
§Only factors associated with a healed rate differential of ⱖ10%, or significant variables, presented.

TABLE 3. Bivariate analysis between selected factors§ and preoperative radiolucency, presented for the pooled phases I-III (n ⫽ 369)
Preoperative Radiolucency
Prognostic Factor
Absent n (%) Present n (%) p Valuea
Preoperative
Gender
Female 100 (50) 101 (50) 0.007
Male 60 (36) 108 (64)
Signs and symptoms
Absent 73 (51) 69 (49) 0.014
Present 87 (38) 140 (62)
Pulp status
Responsive 104 (78) 29 (22) < 0.001
Nonresponsive 56 (24) 180 (76)
Intraoperative
Treatment sessions
1 47 (70) 20 (30) < 0.001
ⱖ2 113 (37) 189 (63)
Root-filling voids
Absent 144 (46) 167 (54) 0.008
Present 16 (28) 42 (72)
Sealer extrusion
Absent 87 (50) 86 (50) 0.012
Present 73 (37) 123 (63)
Bold type face highlights statistical significance.
§Only factors significantly associated with preoperative radiolucency presented.
a␹2 test.

302 Marquis et al. JOE — Volume 32, Number 4, April 2006

 Clinical Research
TABLE 4. Logistic regression model of the outcome of initial endodontic treatment 4 to 6 years after treatment for the pooled phases I-III (n ⫽ 369)
Odds Ratio
Predictor 95% Confidence Interval p Value
(for disease)
Preoperative
Radiolucency 3.55 1.75-7.23 <0.001
(0 ⫽ absent, 1 ⫽ present)
No. of roots 2.17 1.00-4.69 0.050
(0 ⫽ single, 1 ⫽ multiple)
Intraoperative
Complications 2.23 1.10-4.52 0.026
(0 ⫽ absent, 1 ⫽ present)
Bold type face highlights statistical significance.

Pooled Phases I, II, and III Sample were associated with healed rate differentials of ⱖ10% that were not
The pooled examined sample included 373 teeth (50% recall rate) statistically significant. The three variables were further assessed in a
in 325 patients. Table 1 compares the examined study sample to the stratified multivariate analysis.
inception cohort (1370 teeth in 1151 patients). Response bias analysis Stratified multivariate analysis performed on teeth with preopera-
(not shown) confirmed that the attending and lost-to-follow-up popu- tive radiolucency (Table 6) revealed an increased risk of persistent
lations differed significantly only with regards to age. disease for the SBLC technique (OR ⫽ 2.83; CI 1.31-6.13) and com-
Of the pooled sample, 317/373 teeth (85%) were classified as plications (OR ⫽ 2.67; CI 1.13-6.32).
healed and 56/373 (15%) as having disease. Four teeth (one healed and
three diseased) were found to be fractured; as they could potentially
confound the investigation of other variables, they were excluded from Discussion
further analysis reducing the analyzed sample to 369 teeth. Of the 316 Methodology
teeth classified as healed (one fractured tooth excluded), 12 teeth (4%) Clinical outcome research is intended to support decision-mak-
presented with a slight tenderness to percussion. Of the 53 teeth clas- ing, such as selecting between tooth retention via endodontic treatment
sified as having disease (three fractured teeth excluded), five teeth (9%) or extraction and replacement. Because the evidence base for current
presented signs and symptoms (three had pain, one had a sinus tract, endodontic treatment outcome is limited to only few studies that con-
and one had pain and swelling), of which three had a PAI score ⱕ2. form to an adequate level of evidence (13–22, 24), additional studies
Thus, in total, 352/369 teeth (95%) (304 healed and 48 having disease) may add important information to the existing knowledge. On this
were fully functional, without signs, symptoms, or tenderness to per- premise, the Toronto Study project was established in 1993 as a pro-
cussion. In 49 teeth with PAI ⱖ 3, the lesion was smaller than preop- spective, modular observational cohort study designed to assess the 4-
eratively in 21 teeth (43%), unchanged in 10 teeth (20%), and in- to 6-year outcome of endodontic treatment. The first two phases of the
creased or new in 18 teeth (37%). project have been summarized in reports on the outcome of initial
The bivariate analysis (Table 2) suggested only four statistically endodontic treatment (20, 21), orthograde retreatment (29), and api-
significant associations, with a higher healed rate for teeth: (a) in fe- cal surgery (30).
males than males, (b) without than with preoperative radiolucency, (c) The present study assessed the outcome of initial endodontic treat-
single-rooted than multirooted, and (d) without than with intraopera- ment in phase III of the Toronto Study project. The methodology and
tive complications. All other variables were associated with healed rate protocol of this study were consistent with those of the previous phases
differentials of ⬍10%. A further analysis (Table 3) revealed that the (20, 21) except for a few updates to comply with current endodontic
variable “gender” was significantly associated with the variable “preop- techniques. Furthermore, the univariate analysis suggested that the
erative radiolucency,” both in the study sample (p ⫽ 0.007) and in the phase III study sample was comparable to those of phases I and II in
inception cohort (p ⫽ 0.026, not shown). Complications, observed in 63 regard to size and frequencies. The uniformity allowed the pooling of
teeth (17%), comprised a variety of preoperative complexities, including the study samples from all three phases, previously suggested as a ben-
aberrant anatomy (20 teeth) and crack observed in the pulp chamber (3 efit of the Toronto Study design (21). The increased statistical power
teeth), as well as intraoperative complications, including perforation (18 was expected to facilitate the assessment of outcome associations with
teeth), file breakage (11 teeth) and apparently calcified canals that could variables, particularly in the stratified analyses for which the sample size
not be negotiated (14 teeth). The four variables were considered potentially was roughly half that of the entire study sample. Indeed, in the preop-
important and further assessed in the multivariate analysis. erative presence of AP, the outcome was significantly associated with
Multivariate analysis performed on the pooled study sample (Ta- only one variable (number of roots) in the phase I study (20), two
ble 4) revealed an increased risk of persistent disease for preoperative variables (number of roots, treatment technique) in the pooled phases
radiolucency (OR ⫽ 3.55; CI 1.75-7.23), multirooted teeth (OR ⫽ I and II (21), and three variables (number of roots, treatment tech-
2.17; CI 1.00-4.69), and intraoperative complications (OR ⫽ 2.23; CI nique, and complications) as shown herein for the pooled sample of
1.10-4.52). phases I through III. Clearly, a large sample size is essential for assess-
Of 160 teeth without preoperative radiolucency, 149 (93%) ment of outcome predictors in a multifactorial disease process such as
healed. Stratified analysis did not show any significant associations, and apical periodontitis.
all healed rate differentials were ⬍10%. Of 209 teeth treated with ra- The study methodology has already been discussed in detail in the
diolucency present, 167 (80%) healed. Stratified analysis (Table 5) previous reports (20, 21, 29, 30). Arguably, it conformed to criteria
suggested only three statistically significant associations, with a higher defining an adequate level of evidence in regards to the study cohort,
healed rate for: (a) single-rooted than multirooted teeth, (b) FPVC than intervention, outcome assessment, and analysis, with the exception of
SBLC technique, and (c) teeth without than with intraoperative compli- the low recall rate. Despite the monetary compensation offered and
cations. Two additional variables (tooth type, coronal seal material) numerous attempts to contact discontinuers and to encourage dropouts

JOE — Volume 32, Number 4, April 2006 Treatment Outcome in Endodontics 303
Clinical Research
TABLE 5. Stratified bivariate analysis of associations between selected factors§ and the healed rate in teeth treated with preoperative radiolucency 4 to 6 years after
treatment, presented for phase III (n ⫽ 66) and the pooled phases I-III (n ⫽ 209)
Phase III Phases I-III pooled
Prognostic Factor Healed Healed
n p Value n p Value
(% n) (% n)
Preoperative
Tooth type
Anterior 7 100 0.336* 40 90 0.076
Posterior 59 75 169 76
Tooth location
Maxilla 25 64 0.045
Mandible 41 85
No. of roots
1 62 89 0.039
ⱖ2 147 76
Periodontal defect
No 62 76 0.567*
Yes 4 100
Intraoperative
Treatment sessions
1 3 100 1.000*
ⱖ2 63 76
Technique
SBLC 103 73 0.005
FPVC 96 89
Other* 10
Type of sealer
ZOE 65 78 0.227*
Non-ZOE 1 0
Sealer extrusion
Absent 21 86 0.353*
Present 45 73
Complications
Absent 54 82 0.125* 180 83 0.008
Present 12 58 32 63
Coronal seal material
Temporary† 6 33 0.021* 34 68 0.051
Definitive‡ 60 82 175 82
Postoperative
Restoration at follow-up
Temporary 1 100 1.000*
Definitive 63 78
Bold type face highlights statistical significance.
SBLC ⫽ modified step-back preparation, lateral compaction of gutta-percha; FPVC ⫽ flared preparation, vertical compaction of warm gutta-percha.
§Only factors associated with a healed rate differential of ⱖ10%, or significant variables, presented.
*Fischer’s Exact test, ␹2 test otherwise.
†Cavit, ZOE, IRM.
‡Amalgam, composite resin, glass ionomer cement, crown.

to attend the follow-up examination, the recall rate of 50% in the phase analysis revealed that the proportion of young (ⱕ45 yrs) subjects was
III study was similar to that in the previous two phases (20, 21). The significantly higher in the latter than in the former; however, because
recall rate fell below the guidelines suggested for high level of evidence age was not identified as a significant outcome predictor, the study
(6). Theoretically, examination of the lost-to-follow-up population results were unlikely to be impacted by the low recall rate.
(discontinuers and dropouts) could strongly pull the outcome towards The majority of the study cohort comprised a specific dental
healing or disease, according to this population’s characteristics. The school population that could differ from that treated in private practice
recall bias analysis was performed, therefore, to compare the charac- in regards to demographic characteristics or disease severity (6). Be-
teristics of the study sample and the lost-to-follow-up population. The cause the study cohort did not represent the population at large, the

TABLE 6. Logistic regression model of the outcome of initial endodontic treatment in teeth treated with preoperative radiolucency 4 to 6 years after treatment for
the pooled phases I-III. (n⫽ 209)
Predictor Odds Ratio 95% Confidence Interval p Value
Intraoperative
Technique 2.83 1.31-6.13 0.008
(0 ⫽ Schilder, 1 ⫽ alternative)
Complications 2.67 1.13-6.32 0.026
(0 ⫽ absent, 1 ⫽ present)
Bold type face highlights statistical significance.

304 Marquis et al. JOE — Volume 32, Number 4, April 2006

 Clinical Research
results might not be generalized beyond this specific study cohort. Nev- healing of AP is a dynamic process that requires different lengths of time
ertheless, the Graduate Endodontics Clinic at the Faculty of Dentistry has for different teeth. In a long-term follow-up study (31), about 6% of
functioned as a specialty referral clinic; therefore, the scope of clinical teeth that had persistent AP 10 to 17 yrs after initial treatment were
conditions treated in this study was considered to be comparable to that completely healed a further 10 yrs later. Notably, of the teeth classified
in an endodontic specialty practice. as having disease in the present study, 43% had a smaller lesion than
The close supervision of the treatment providers (graduate stu- preoperatively, possibly suggesting a slower than usual healing process
dents) by qualified endodontists as well as the principal treatment pro- for at least some of these teeth. This consideration emphasizes the need
tocol, were unchanged from the previous phases. Importantly, however, for extended follow-up when a possible healing process is suggested by
engine-driven nickel titanium instruments and surgical operating mi- reduction of the lesion and absence of clinical signs and symptoms
croscopes were introduced during the period when the phase III cohort (21). Possibly, in the long term the difference in outcome in teeth
was treated. The FPVC technique also was modified and adhered less to treated with or without AP may be less than what is apparent in the
the specific guidelines established by Schilder (25, 26). Whether or not shorter term, such as the 4- to 6-yr follow-up reported herein. Secondly,
these changes have influenced the outcome of treatment was expected the poorer outcome may be a result of a small proportion of teeth with
to be gleaned from comparison of the phase III and the previous phases’ AP where the infecting bacteria are not situated in the root canal system
results. The overall 86% healed rate in the phase III study was consistent but rather in the periapical tissues (32, 33) or on the outer surface of
with the 87% rate in phase II (21), and both were higher than the 81% the root (32). Even if root canal treatment can effectively control intra-
rate in phase I (20). Thus, the aforementioned changes in protocol did canal infection in all teeth where such is present, it is unlikely to curtail
not appear to influence the outcome of treatment. extraradicular infection. Thus, even if novel root canal disinfection reg-
imens such as potent irrigants, intracanal medication or lasers are more
Overall Outcome and Main Predictors effective than the conventional ones, and they become widely adopted,
Adding the third phase to the previous two made it apparent that the healing rate is unlikely to match that in teeth treated without AP.
any significant associations suggested by the bivariate analyses of either The number of roots was identified as a significant outcome pre-
single phase, have neither been consistently corroborated by successive dictor, with an overall healed rate differential of 9%. However, the
phases nor substantiated in the pooled samples. Likewise, not all vari- methodology precluded any insight into whether this finding reflected a
ables suggested to be significant by the bivariate analyses of the pooled greater challenge to eliminate root canal infection in multirooted teeth,
samples were confirmed as significant outcome predictors by the mu- or the use of the tooth (as opposed to the root) as the evaluated unit with
tivariate analysis. This inconsistency underscored the limitations of the the risk of persisting AP multiplied by the number of roots in any treated
bivariate analysis, which does not take into account confounding effects tooth (21). Nevertheless, when advising patients of the prognosis for a
when assessing any specific variable, but only suggests potential vari- single-rooted or multirooted tooth, the specific healing rates for either
ables for inclusion in logistic regression models. In view of these con- group can be quoted rather than average figures that do not truly rep-
siderations, only the results in the pooled study sample are discussed resent either group.
below, with greater emphasis placed on the multivariate than the biva- Intraoperative complications had a significant negative impact on
riate analyses. treatment outcome, particularly in teeth with preoperative AP where the
Using both clinical and radiographic outcome measures, the over- healed rate differential reached 20%. This finding was in agreement
all healed rate in the pooled sample was 85%. However, as many as 95% with other studies (8, 10, 13). By their nature, all of perforation, file
of the teeth presenting for follow-up were symptom-free and fully func- breakage, untreated canals, cracks, and aberrant anatomy can either
tional. This finding was within the range of 88% to 97% reported in promote infection or interfere with its elimination; therefore, their im-
other current studies (14, 17–19). Furthermore, if the sole presence of pact on the outcome of treatment was expected, and the differential in
a slight tenderness to percussion was considered as not hindering func- outcome was already apparent in the pooled phases I and II, but without
tion, 99% of the teeth in this study (316 healed and 48 having disease) statistical significance (21). Some of these complications, however, may
would be functional. Inasmuch as scientific rigor requires the use of be much less frequent in the general population treated in private prac-
both clinical and radiographic outcome measures, individuals/patients tice (34) than in this university-based study. Thus the impact of com-
may select to base their treatment preferences exclusively on the plications on the outcome of initial endodontic treatment in the general
chances of eliminating signs and symptoms (7). This concept of patient population might be proportionally lower than observed herein. Impor-
autonomy in clinical decision-making provides the rationale and sup- tantly, without the complications, the healed rate in teeth with preoper-
ports the relevance of reporting treatment outcomes in regards to both ative AP was 83%.
healing and functionality (7). Therefore, patients who are weighing
endodontic treatment against tooth extraction and replacement should Suggested Predictors and Clinical Significance
be advised that, based on this and several other current studies (14, Treatment technique was suggested as a significant outcome pre-
17–19), the chance of endodontically treated teeth to remain fully func- dictor, corroborating the phase II study (21) results, but only in teeth
tional 4 to 6 years after treatment is 88% to 97%, even if some may with preoperative AP. The association between outcome and treatment
present radiographic signs consistent with disease. technique was not confounded by any other outcome predictor (anal-
The multivariate analysis performed on the entire pooled sample ysis not shown). As in the phase II study (21), the healed rate after use
identified three significant outcome predictors, one of which (preop- of FPVC was 16% higher than after use of SBLC. Thus the modification of
erative AP) was previously identified in phases I (20) and II (21) of the the original Schilder technique (25, 26) did not appear to notably affect
Toronto Study project. Preoperative AP had the strongest predictive the outcome. The fact that this finding was corroborated by the multi-
ability, in agreement with most previous studies (8 –10, 12–14, 18, 23, variate analysis in a successive phase of the Toronto Study strengthened
24). Without preoperative AP, the 93% healed rate was in the middle of its validity. Nevertheless, it should be considered as suggestive at best,
the range (88-97%) reported in previous studies (8 –10, 12–14, 18, because this study was only quasirandomized for treatment technique
20 –23). With preoperative AP, the 80% healed rate was in the middle of and other variables were not controlled, as is required for the highest
the range (73-90%) reported in previous studies (8 –11, 13–23). Cer- level of evidence (5, 6). In the absence of other studies that compare
tain aspects of treatment in teeth with AP require consideration. Firstly, FPVC and SBLC, the results of this study underlined the need for ran-

JOE — Volume 32, Number 4, April 2006 Treatment Outcome in Endodontics 305
Clinical Research
domized controlled trials designed specifically to compare the outcome 9. Engström B, Hard AF, Segerstad L, Ramström G. Frostell G. Correlation of positive
of treatment associated with these two treatment techniques. cultures with the prognosis for root canal treatment. Odontol Revy 1964;15:257–70.
10. Kerekes K, Tronstad L. Long-term results of endodontic treatment performed with a
Although the sample has grown 3-fold since the phase I study (20),
standardized technique. J Endod 1979;5:83–90.
none of the variables significantly influenced the outcome in teeth with- 11. Byström A, Happonen RP, Sjögren U, Sundqvist G. Healing of periapical lesions of
out preoperative AP. On the contrary, several large healed rate differ- pulpless teeth after endodontic treatment with controlled asepsis. Endod Dent Trau-
entials of ⱖ10%, highlighted in the phase I and II studies (20, 21), have matol 1987;3:58 – 63.
diminished below 10% when the three phases were pooled. These find- 12. Ørstavik D, Kerekes K, Eriksen HM. Clinical performance of three endodontic sealers.
ings suggested that with such small differences in outcome among teeth Endod Dent Traumatol 1987;3:178 – 86.
13. Sjögren U, Hagglund B, Sundqvist G, Wing K. Factors affecting the long-term results of
without preoperative AP, a much larger sample size would be required
endodontic treatment. J Endod 1990;16:498 –504.
to identify significant outcome predictors. 14. Ørstavik D. Time-course and risk analyses of the development and healing of chronic
As the study sample of the Toronto Study project has increased with apical periodontitis in man. Int Endod J 1996;29:150 –5.
more phases added, the multifactorial influences on the outcome of 15. Sjögren U, Figdor D, Persson S, Sundqvist G. Influence of infection at the time of root
intitial endodontic treatment have become more apparent. Clinicians filling on the outcome of endodontic treatment of teeth with apical periodontitis. Int
should take notice of the variables identified as significant outcome Endod J 1997;30:297–306.
16. Trope M, Delano EO, Ørstavik D. Endodontic treatment of teeth with apical periodon-
predictors. The importance of the preoperative ones (apical periodon- titis: single vs. multivisit treatment. J Endod 1999;25:345–50.
titis and number of roots) is that they can be recognized at the stage 17. Weiger R, Rosendahl R, Löst C. Influence of calcium hydroxide intracanal dressings
when prognosis is projected and the patient is weighing treatment of the on the prognosis of teeth with endodontically induced periapical lesions. Int Endod
tooth against extraction. The importance of the intraoperative ones J 2000;3:219 –26.
(complications and possibly treatment technique) is that clinicians 18. Hoskinson SE, Ng YL, Hoskinson AE, Moles DR, Gulabivala K. A retrospective com-
might be able to control the outcome of treatment by preventing pro- parison of outcome of root canal treatment using two different protocols. Oral Surg
Oral Med Oral Pathol Oral Radiol Endod 2002;93:705–15.
cedural errors and possibly, by making informed choices of treatment 19. Peters LB, Wesselink PR. Periapical healing of endodontically treated teeth in one and
techniques. The authors suggest that as more knowledge becomes avail- two visits obturated in the presence or absence of detectable microorganisms. Int
able on the outcome predictors in endodontic treatment, the greater Endod J 2002;35:660 –7.
becomes the onus on the clinicians to apply this knowledge when pro- 20. Friedman S, Abitbol S, Lawrence HP. Treatment outcome in endodontics: the Toronto
viding endodontic care to their patients. Study. Phase 1: Initial treatment. J Endod 2003;29:787–93.
In conclusion, in phase III of the Toronto Study project, 113/132 21. Farzaneh M, Abitbol S, Lawrence HP, Friedman S. Treatment outcome in endodon-
tics: the Toronto Study. Phase II: Initial treatment. J Endod 2004;30:302–9.
teeth (86%) were classified as healed. The addition of phase III dem- 22. Huumonen S, Lenander-Lumikari M, Sigurdsson A, Ørstavik D. Healing of apical
onstrated the limitations of the bivariate analyses in identifying outcome periodontitis after endodontic treatment: a comparison between a silicone-based and
predictors in endodontic treatment. After combining the data collected a zinc oxide-eugenol-based sealer. Int Endod J 2003;36:296 –301.
from phase III with that collected in the two previous phases, the mul- 23. Peters OA, Barbakow F, Peters CI. An analysis of endodontic treatment with three
tivariate analysis identified preoperative radiolucency, number of roots, nickel-titanium rotary root canal preparation techniques. Int Endod J
and intraoperative complications as significant outcome predictors in 2004;37:849 –59.
24. Marending M, Peters OA, Zehnder M. Factors affecting the outcome of orthograde
initial endodontic treatment. The suggested outcome predicting ability root canal therapy in a general dentistry hospital practice. Oral Surg Oral Med Oral
of treatment technique in teeth with preoperative apical periodontitis Pathol Oral Radiol Endod 2005;99:119 –24.
requires confirmation by results of appropriately designed randomized 25. Schilder H. Filling root canals in three dimensions. Dent Clin North Am 1967;
controlled trials. 723– 44.
26. Schilder H. Cleaning and shaping the root canal. Dent Clin North Am 1974:269 –96.
References 27. Ørstavik D, Kerekes K, Eriksen HM. The periapical index: a scoring system for ra-
diographic assessment of apical periodontitis. Endod Dent Traumatol
1. Curtis DA, Lacy A, Chu R, Richards D, Plesh O, Kasrovi P , et al. Treatment planning 1986;2:20 –34.
in the 21st century: what’s new? J Calif Dent Assoc 2002;30:503–10. 28. Landis JR, Koch GG. The measurement of observer agreement for categorical data.
2. Matosian GS. Treatment planning for the future: endodontics, post and core, and Biometrics 1977;33:159 –74.
periodontal surgery– or an implant? J Calif Dent Assoc 2003;31:323–5.
29. Farzaneh M, Abitbol S, Friedman S. Treatment outcome in endodontics: the Toronto
3. Somborac M. Implant treatment versus endodontic retreatment: a contemporary
study. Phases I and II: Orthograde retreatment. J Endod 2004;30:627–33.
dilemma. Oral Health 2003;93:8 –15.
4. Friedman S. Prognosis of initial endodontic therapy. Endod Topics 2002;1:59 – 88. 30. Wang N, Knight K, Dao T, Friedman S. Treatment outcome in endodontics: The
5. Department of Clinical Epidemiology and Biostatistics, McMaster University. How to Toronto Study. Phases I and II: Apical surgery. J Endod 2004;30:751– 61.
read clinical journals: III. To learn the clinical course and prognosis of disease. Can 31. Molven O, Halse A, Fristad I, MacDonald-Jankowski D. Periapical changes following
Med Assoc J 1981;124:869 –72. root-canal treatment observed 20 –27 years postoperatively. Int Endod J
6. Sackett D, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology: A Basic Science 2002;35:784 –90.
of Clinical Medicine, 2nd ed. Boston: Little, Brown; 1991. 32. Tronstad L, Sunde PT. The evolving new understanding of endodontic infections.
7. Friedman S, Mor C. The success of endodontic therapy: healing and functionality. J Endod Topics 2003;6:57–77.
Calif Dent Assoc 2004;32:493–503. 33. Siqueira JF. Periapical actinomycosis and infection with Proprionibacterium pro-
8. Strindberg LZ. The dependence of the results of pulp therapy on certain factors. An pionicum. Endod Topics 2003;6:78 –95.
analytic study based on radiographic and clinical follow-up examination. Acta Od- 34. Vire DE. Failure of endodontically treated teeth: classification and evaluation. J Endod
ontol Scand 1956;14(Suppl 21). 1991;17:338 – 42.

306 Marquis et al. JOE — Volume 32, Number 4, April 2006