ORIGINAL ARTICLE

Retrospective analysis of long-term stable and

unstable orthodontic treatment outcomes
Jonathan Perry Ormiston,a Greg J. Huang,b Robert M. Little,c Jay D. Decker,d and Geoffrey D. Seuke
Puyallup and Seattle, Wash
Introduction: The purpose of this study was to compare groups of patients with the most stable and the
most unstable treatment results as rated by the peer assessment rating (PAR) index to identify factors
associated with stability. All factors with significant crude odds ratios were investigated to create a multiple
logistic regression model that could be used to predict stability. Methods: The sample of 86 patients (30
male, 56 female), from the postretention archives at the University of Washington, was not restricted to
specific malocclusion types or treatment modalities with the exception of Angle Class III patients, who were
excluded. The sample was divided into 2 groups, stable (n 45) and unstable (n 41), based on
postretention unweighted PAR scores and PAR score changes between posttreatment and postretention.
Model and radiographic measurements were made before treatment, after treatment, and after retention
(average 14.4 years). Results: The results showed that male sex and a sustained period of growth were
related, and both were associated with increased instability. The initial severity of malocclusion, as graded
by the PAR index and the irregularity index, was negatively correlated with postretention stabilityie,
patients with more severe index scores before treatment tended to be less stable. Differences in American
Board of Orthodontics scores after treatment were diminished after retention. Conclusions: The factors
associated with predicting stability were pretreatment arch length, pretreatment PAR score, molar classification, and sex. (Am J Orthod Dentofacial Orthop 2005;128:568-74)

aintaining teeth in their corrected positions

after orthodontic treatment is challenging,
and practitioners commonly recommend
long-term or permanent retention. Prior long-term retention studies have often selected subjects based on a
particular therapy (eg, first premolar extraction) and
then followed them over time. However, because longterm stability is rare, these studies usually had an
inadequate number of stable cases for statistical analyses. An alternative strategy is to identify stable and
unstable subjects from a defined population, and then
compare their pretreatment and posttreatment records.
This allows the investigator to control the numbers of
stable and unstable subjects in the study, leading to
increased statistical power. Additionally, if logistic
regression is used, many risk factors for stability can be

Private practice, Puyallup, Wash.

Associate professor, Department of Orthodontics, University of Washington,
Seattle.
c
Professor, Department of Orthodontics, University of Washington, Seattle.
d
Affiliate associate professor, Department of Orthodontics, University of
Washington, Seattle.
e
Dental student, University of Washington, Seattle.
Funded by the University of Washington Orthodontic Memorial Fund
Reprint requests to: Greg J. Huang, Associate professor, Box 357446,
University of Washington, Seattle, WA 98195-7446; e-mail, ghuang@u.
washington.edu.
Submitted, May 2004; revised and accepted, July 2004.
0889-5406/$30.00
Copyright 2005 by the American Association of Orthodontists.
doi:10.1016/j.ajodo.2004.07.047
b

568

analyzed simultaneously, and adjustments can be made

for confounding variables. The purpose of this study
was to compare groups of patients with the most stable
and the most unstable long-term occlusal results, as
rated by the peer assessment rating (PAR) index, to
identify factors associated with stability.
Blake and Bibby1 suggested that many factors
might be related to postretention crowding. Their review focused on 15 potential factors of stability. Several of these merit additional investigationlong-term
growth, pretreatment occlusal characteristics, and posttreatment occlusal characteristics.
Isaacson et al2 found that, when growth at the
condyles is greater than that in the mid face, forward
rotation of the mandible occurs, resulting in anterior
movement of the mandible. Many researchers have
postulated that this growth of the mandible is related to
long-term crowding via several possible mechanisms.3-7 Other researchers have found no significant
correlation between continued growth and the development of long-term crowding,8,9 and some have reported
that continued growth might have a positive effect on
stability.10-12 The effects of growth are not easily
described, and their effects on the occlusion could be
positive, negative, or neutral.
Pretreatment occlusal relationships might also influence stability. Many studies have shown that patterns
of relapse tend to be toward a patients original dental

American Journal of Orthodontics and Dentofacial Orthopedics

Volume 128, Number 5

relationship.12-15 In contrast to this trend, Little et al16

found that nearly half of the rotations and displacements analyzed relapsed to positions that were different
from the starting malocclusion. Additionally, although
some studies have attempted to find pretreatment factors associated with postretention crowding, most attempts have failed to show any reliable predictors of
stability.9,17 Thus, controversy exists about the longterm impact of the pretreatment condition on the
postretention conditions.
Although many researchers have advocated treating
orthodontic patients to very high standards to increase
long-term stability,18-20 several recent studies have
found no correlation between low posttreatment PAR
scores (representing a high-quality finish) and longterm stability.16,21-24 Again, there is disagreement on
how this variable affects long-term stability.
MATERIAL AND METHODS

The sample was selected from approximately 600

postretention cases in the University of Washington
archives. The unweighted PAR25 was used to define
stable and unstable patients for this retrospective casecontrol study. Those with unweighted PAR scores
under 3 at postretention were placed in the stable group,
and those with unweighted PAR scores over 10 were
placed in the unstable group. Other inclusion criteria
were the availability of patient casts, lateral cephalometric radiographs, and clinical records that described
sex, age at start of treatment, length of treatment, and
length of retention. The casts and radiographs were
from 3 different time points: pretreatment (T1), posttreatment (T2), and a minimum of 89 months (7.4
years) postretention (T3). No patient was categorized as
unstable unless there was an increase of 5 or more
unweighted PAR points between T2 and T3, and no
patient was categorized as stable if there was a decrease
of more than 5 unweighted PAR points between T2
and T3.
The PAR and irregularity index (II) were used to
grade the patient casts at T1. The PAR, II, and cast
components of the American Board of Orthodontics
(ABO) scoring system were used to grade the casts at
T2 and T3. (The ABO system was not used at T1
because it was developed for evaluating posttreatment
occlusal relationships. Also, the interproximal contact
category of the ABO was not included for study due to
the many subjects who were fully banded.)
In addition to the indexes listed above, measurements of arch length (from the mesial contact point of
the first molars to the contact point between the central
incisors), intermolar width, intercanine width, and mandibular anterior tooth width were made at all 3 time

Ormiston et al 569

points. To prevent bias, the models were measured

randomly in blocks of 10; all T3 measurements were
made first, followed by T2 and then T1. Graders were
blinded to the group (stable or unstable) to which the
models belonged.
Molar relationships were assessed on the models,
by using the location of the mesiobuccal cusp of the
maxillary first molar and the buccal groove of the
mandibular first molar as reference points. Any cast
with maxillary buccal cusps on either the right or left
side that were 3 mm or more from the buccal groove of
the mandibular molar in the Class II direction was
classified as Class II. Any casts with differences of 3
mm or more in the Class III direction were considered
Class III. Casts with left and right maxillary mesiobuccal cusps within 3 mm of the mandibular molar buccal
groove were considered Class I.
Investigators were trained and calibrated for each of
the occlusal indexes used in this study. All PAR and II
measurements were made by one author (G.D.S.), and
all ABO measurements were made by another author
(J.P.O.). Intraexaminer reliability was assessed by remeasuring 20 randomly selected cases. Lateral cephalometric radiographs were used to assess maxillary and
mandibular growth. Nine points on the cephalometric
radiograph were digitized (sella, A point, B point,
nasion, ANS [where it is 3 mm thick], articulare,
pogonion, gnathion, and menton). A modified Harvold
analysis was used (articulare was substituted for condylion) comparing maxillary and mandibular unit
length at the 3 time intervals.26 The maxillary unit was
measured from articulare to a point where the anterior
nasal spine was 3 mm thick. Several subjects had
cephalometric radiographs with poorly defined A
points. In these cases, a maxillary template was constructed from a radiograph that had a well-defined A
point. Information about age at start of treatment, time
in treatment, time in retention, postretention interval,
type of retainer used, and teeth extracted was obtained
from the University of Washington postretention
records database.
T tests were used to assess differences between the
stable and unstable groups for continuous variables,
with a P value of .05 to determine statistical significance. Chi-square tests were used to test for differences
between the stable and unstable groups for categorical
variables, with a P value of .05 for statistical significance. Analysis of variance (ANOVA) was used to
compare the ABO score categories at T3 (the Bonferroni adjustment was used for multiple comparisons).
Crude odds ratios were calculated for each factor
thought to be related to stability and for all potentially
confounding variables. Factors that were associated

570 Ormiston et al

Table I.

American Journal of Orthodontics and Dentofacial Orthopedics

November 2005

Demographics
Unstable (n 41)

Variable
Mean age at T1 (y)
Mean treatment time (y)
Mean retention time (y)
Mean post retention time (y)
Mean T2 MN 3-3 width (mm)
Mean T3 MN 3-3 width (mm)
Angle class (% Class II)
Class II Division 1 (% of all Class II)
Sex (% female)
Extraction (% with extractions)

Stable (n 45)

P value

Mean or %

SD

Minimum

Maximum

Mean or %

SD

Minimum

Maximum

.276
.325
.242
.182

12.6
2.3
2.5
15.1
36.7
36.4
70.7
72.4
48.8
63.4

2.4
.9
1.6
5.0
3.8
3.8

8.8
.9
.5
9.3
24.9
24.6

23.9
5.3
8.4
32.8
42.2
42.0

12.1
2.1
3.0
13.8
36.0
35.6
46.7
76.2
80.0
51.1

2.3
1.1
1.8
3.7
3.9
4.0

7.2
.8
.5
7.4
21.5
21.6

16.2
7.0
10.1
23.3
44.9
44.0

.596
.024
.764
.002
.250

MN 3-3 width, summed width of mandibular teeth.

with stability in the univariate analyses were investigated in multiple logistic regression analyses.
RESULTS

A total of 88 patients met the inclusion criteria.

Only 2 had Class III malocclusions, and it was decided
to exclude them from the sample. Of the remaining 86
subjects, 30 were male and 56 female. The stable group
consisted of 45 subjects and the unstable group consisted of 41 subjects (Table I). Sex differences between
the groups were statistically significant, with the stable
group having significantly more females than the unstable group. The differences between the stable and
unstable groups for the following variables were not
significant: age at the start of treatment, treatment time,
retention time, postretention time, and percentages with
extractions, extraction patterns, and type of retainer.
There were significantly more Class II patients in
the unstable group than in the stable group (Table I). Of
the Class II patients, there were no statistically significant differences between the 2 groups for Angle Class
II divisions or subdivisions at T1. The widths of the
mandibular anterior teeth at T2 and T3 were compared
to determine whether reproximation had occurred during the retention period (Table I). The difference of
0.14 mm was not significant.
The mean ABO score decreased between T2 and T3
in the stable group (improvement in occlusion), but it
increased in the unstable group (deterioration in occlusion) (Table II). The differences in ABO scores between the stable and unstable groups were significant at
both T2 and T3. The unstable group had significantly
higher scores at both time points (Table II).
To further investigate the impact of posttreatment
occlusion, the subjects were divided into 4 categories
based only on ABO scores at T2 (group 1, scores of
10-20; group 2, scores of 20-30; group 3, scores of

30-40; group 4, scores over 40). The mean scores for

groups 1, 2, and 4 did not show statistically significant
changes between T2 and T3. Group 3 had a statistically
significant decrease between T2 and T3 (Table III).
There were significant differences in the weighted
and unweighted PAR scores at all 3 time points (Table
II), with the stable group consistently having lower
scores. The stable group started with a significantly
lower mean II score than the unstable group. The
groups were finished to a similar level. At T3, both
groups had increases in II scores, but the unstable group
had significantly higher scores than the stable group
(Table II).
During treatment, the increases in mean intercanine
width for both groups were not statistically different
(Table II). From T2 to T3, both groups had decreases in
intercanine width to distances below those at T1. The
unstable group had a significantly larger T2-T3 decrease than the stable group (Table II). There were no
significant differences between the stable and unstable
intermolar widths at any time (Table II).
The differences in arch length between the stable
and unstable groups were significant at T1; the stable
group started slightly greater than the unstable group
(Table II). Arch-length measurements continually decreased for both groups, but the differences were not
significant at either T2 or T3.
At T1 and T2, the difference in overbite measurements was not significantly different between the 2
groups. At T3, the unstable group had a significantly
greater overbite measurement than the stable group
(Table II). Changes in overjet followed similar trends.
The unstable group started with significantly more
overjet and then decreased to a level that was not
statistically different from the stable group. At T3,
overjet in the unstable group increased to a level that
was significantly larger than that of the stable group

Ormiston et al 571

American Journal of Orthodontics and Dentofacial Orthopedics

Volume 128, Number 5

Table II.

Mean values at T1, T2, and T3

Unstable
Variable

Time

P value

ABO score
ABO score
Unweighted PAR
Unweighted PAR
Unweighted PAR
Weighted PAR
Weighted PAR
Weighted PAR
II
II
II
Intercanine width (mm)
Intercanine width (mm)
Intercanine Width (mm)
Intermolar width (mm)
Intermolar width (mm)
Intermolar width (mm)
Arch length (mm)
Arch length (mm)
Arch length (mm)
Overbite (mm)
Overbite (mm)
Overbite (mm)
Overjet (mm)
Overjet (mm)
Overjet (mm)
Mean molar distance from Class I (mm)
Mean molar distance from Class I (mm)
Mandibular plane angle ()
Mandibular plane angle ()
Mandibular plane angle ()
SNA ()
SNA ()
SNA ()
SNB ()
SNB ()
SNB ()
ANB ()
ANB ()
ANB ()
Maxillary Harvold (mm)
Maxillary Harvold (mm)
Maxillary Harvold (mm)
Mandibular Harvold (mm)
Mandibular Harvold (mm)
Mandibular Harvold (mm)

T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3
T1
T2
T3

.0072
.0000
.0000
.0000
.0000
.0001
.0004
.0000
.0231
.0965
.0000
.8092
.0649
.9503
.3008
.9083
.9899
.0263
.4667
.0721
.1744
.7377
.0036
.0299
.0523
.0000
.9430
.0140
.9548
.6342
.7774
.5236
.5740
.5245
.0451
.0349
.0190
.0325
.0199
.0155
.9162
.9911
.3532
.3470
.4669
.6347

Mean

SD

31.6 12.3
35.9 12.0
18.2
5.3
3.7
2.1
12.6
1.7
35.8
8.9
5.5
4.1
21.6
5.1
5.7
3.7
1.2
1.0
5.3
2.2
25.0
2.4
26.7
3.1
24.6
2.9
41.8
3.3
42.1
2.5
41.8
3.1
57.4
4.8
53.6
5.8
50.3
6.3
4.5
2.3
2.4
1.1
3.6
1.9
6.5
3.7
2.5
0.9
3.93 1.50
0.37 1.31
1.23 1.64
35.7
6.6
36.1
6.8
34.2
7.6
81.6
3.4
80.1
3.4
80.5
4.0
76.0
3.5
76.0
3.6
76.5
3.9
5.6
2.0
4.1
2.3
4.0
2.8
86.4
5.5
87.8
5.6
90.6
6.4
99.4
4.8
104.5
6.3
109.4
8.7

(Table II). When the sample was stratified based on the

T1 molar relationship measurements, the unstable
group was found to have started significantly more
Class II than the stable group (Table II). There were no
significant differences between the stable and unstable
groups for the percentages of Class II division or
subdivisions types (Table I).
There were no significant differences between the

Stable

Minimum

Maximum

10.0
20.0
9.0
1.0
11.0
21.0
1.0
13.0
0.5
0.0
0.0
19.2
17.4
15.1
33.6
37.0
36.5
45.5
45.9
38.0
1.0
0.0
1.5
2.0
0.0
1.00
7.00
5.75
24.1
25.1
21.6
73.0
72.2
70.8
67.9
68.2
69.6
2.1
0.2
1.7
70.1
72.5
74.5
87.7
93.2
95.6

66.0
64.0
29.0
9.0
17.0
58.0
17.0
41.0
15.6
4.6
9.1
29.4
36.1
29.3
48.8
46.2
48.8
68.0
64.7
62.5
10.0
4.0
9.0
13.5
5.0
8.50
1.75
2.75
50.1
52.5
53.8
89.5
86.0
87.4
82.0
81.9
83.7
11.2
11.8
12.0
95.0
97.6
100.9
110.0
121.6
132.2

Mean

SD

25.2
9.0
17.8
6.2
13.3
5.0
2.0
1.3
1.5
0.7
27.6 10.2
2.9
2.3
2.5
1.9
3.9
3.6
0.9
0.7
1.7
0.8
24.7
2.2
25.6
2.5
24.6
3.0
42.5
2.6
42.2
2.8
41.8
3.2
59.7
4.8
54.5
6.2
52.6
5.8
3.9
2.0
2.3
1.1
2.7
0.9
4.9
3.5
2.1
0.7
2.42 0.74
0.39 1.17
0.34 0.62
35.6
5.1
35.5
5.4
33.8
5.9
82.2
4.3
80.6
4.3
81.0
4.5
77.6
3.8
77.7
3.7
78.4
3.6
4.6
2.6
2.9
2.5
2.6
2.4
86.2
5.2
87.8
4.7
89.4
5.7
100.5
6.5
105.5
6.3
108.6
7.1

Minimum

Maximum

10.0
5.0
2.0
0.0
0.0
3.0
0.0
0.0
0.0
0.0
0.0
16.5
15.1
14.3
37.3
36.7
37.2
47.4
44.0
43.8
0.5
1.0
1.0
2.0
0.0
1.5
7.0
4.0
23.9
22.0
21.7
74.2
72.6
72.9
69.7
70.0
71.6
1.6
2.7
1.8
75.2
79.2
80.4
88.9
95.1
96.8

51.0
34.0
27.0
6.0
2.0
46.0
9.0
8.0
14.7
2.7
3.4
30.0
30.4
32.3
48.6
49.1
48.9
69.9
64.0
61.9
13.5
5.0
4.0
12.0
4.0
5.0
1.5
1.8
45.2
44.3
45.2
91.0
91.1
95.2
87.3
88.8
88.5
9.9
9.1
9.0
99.7
99.8
101.0
118.0
122.6
126.2

stable and unstable groups in mandibular plane or SNA

measurements at any time (Table II). At T1, the
differences between the stable and unstable groups
were not significant for either the maxillary or the
mandibular Harvold measurements (Table II), nor were
the differences in maxillary growth between T1 and T2.
From T2 to T3, the unstable group experienced significantly more maxillary growth than the stable group

572 Ormiston et al

American Journal of Orthodontics and Dentofacial Orthopedics

November 2005

Table III.

ABO score categories

Category

T2 point range

T2 mean

T3 mean

P value

10-20
20-30
30-40
over 40

15.1
24.1
34.1
47.9

19.8
24.8
26.8
40.6

.0679
.8778
.0012
.1248

19
32
23
12

1
2
3
4

(Table II). The unstable group had more mandibular

unit growth from T1 to T2 and from T2 to T3, but the
difference in growth was not statistically significant for
either time (Table II).
All factors with significant crude odds ratios were
investigated to create a multiple logistic regression
model that could be used to predict stability. The
variables in the final model were sex, T1 PAR score,
molar classification, and arch length. The predictive
formula was determined to be:
Pinstability
e[3.63(1.49Sex)(0.13PAR1)(0.57Mol)(0.12AL1)]
1 e[3.63(1.49Sex)(0.13PAR1)(0.57Mol)(0.12AL1)]
where PAR1 is initial unweighted PAR score, sex is
coded as 0 (female) or 1 (male), Mol is molar classification (1 Class I, 2 Class II), and AL1 is initial
arch length, e represents the natural logarithmic base,
and 3.63 is a constant from the regression analysis.
In the intrarater reliability study, all linear cast
measurements had a root mean square error of less than
1 mm, and all angular radiographic measurements had
a root mean square error of less than 1. The root mean
square error of all PAR measurements was under 2
PAR points with the exception of the T1 weighted
PAR, which was just over 4 PAR points. The root mean
square error for the ABO was less than 2 points.
DISCUSSION

The predictive model should be discussed, because

it encompasses the variables that related to long-term
stability in this adolescent population. The first variable
is sex, which has an adjusted odds ratio of 4.4,
indicating that males are more than 4 times as likely as
females to have unstable occlusions, with all other
variables in the model constant. Because male sex was
closely related to increased facial growth, this variable
indirectly implicates growth as being related to instability in this sample. However, there are other potential
explanations to this finding, such as possible sex
differences in compliance during treatment or retention.
The tendency for males to grow later than females,
as seen in this study, agrees with the observations of
Sinclair and Little8 in studying the growth of untreated

normals. However, the results of our study differ from

those of previous studies that have found no relationship between growth and stability.9 Sinclair and Little,27 in studying untreated normal subjects who were
last assessed about 20 years of age, found women to
have more deterioration of occlusal relationships than
men. However, our study primarily used the comprehensive PAR index, whereas several other studies used
individual measures to grade patient casts. There were
also differences in the samples in these projects. Our
study used a combination of Angle classifications and
extraction patterns, while others might have used a
single Angle classification or specific extraction patterns. Additionally, the use of 2 extreme groups based
on PAR scores at T3 might have enhanced the ability of
our study to detect differences between the stable and
unstable groups.
Our study supports the results of Schudy3 and
Vaden et al28 in finding relationships between growth
and stability. Several other studies highlight the relationship of growth to the correction of malocclusions.10-12 The results of our study, combined with
several studies listed above, suggest that growth is a
factor in stability. It is apparent that growth can correct
poor occlusal relationships and cause good relationships to degenerate. Whether growth will help or hinder
a patient depends primarily on the initial malocclusion,
the posttreatment occlusion, and the amount and direction of subsequent growth. Schudy3 suggested that
retention should be used until the end of growth. This
might result in retaining males for longer periods than
females.
The next 3 variables in the predictive model all
relate to the initial occlusal condition. The adjusted
odds ratio for the initial PAR score was 1.14 per PAR
point difference. A clinically meaningful PAR difference might be about 5 points, so the formula predicts,
for every 5-point increase in initial PAR score, that a
subjects chances to be unstable almost double, with all
other factors constant (1.145 1.84). The next variable
is the initial molar classification. The adjusted odds
ratio for this variable is 1.77. Although this factor was
not significant in the final model, it was included to
allow adjustment for molar classification. Thus, the
odds ratios for the other variables in the model compare
between subjects with the same molar classification.
The trend suggested by this increased odds ratio is that
Class II subjects are about twice as likely to be
unstable, with all other factors constant. The final
variable in the predictive model is initial arch length.
Each additional millimeter of pretreatment arch length
increases the odds for stability about 12%. Thus, a
patient with 5 mm more of arch length before treatment

American Journal of Orthodontics and Dentofacial Orthopedics

Volume 128, Number 5

would have approximately half the risk of being unstable (0.885 0.53). This variable intuitively makes
sense, because the greater the initial arch length, the
more likely that a patient has adequate or excess space.
If we expect a return to the original condition, these
patients are less likely to have crowding in the long
term.
Birkeland et al15 and Simons and Joondeph12 also
found that patients with more extreme malocclusions
before treatment tended to have greater relapses postretention. Shields et al,9 in contrast to our study, found
no predictors of stability. The difference between our
study and that of Shields et al might be the differences
in sample selection and scoring discussed earlier. Fidler
et al17 were also unable to detect any predictors of
stability. This might be because few subjects experienced significant amounts of relapse, which might have
led to reduced statistical power and the failure to detect
any predictors of relapse. The II at T1 was also
predictive of stability, having a crude odds ratio that
was similar to that of the PAR score at T1. Kahl-Nieke
et al29 also found that some pretreatment variables were
associated with postretention crowding and increases in
incisor irregularity. Their study involved a population
that was primarily treated with removable appliances,
and, therefore, the results might not completely apply to
a population treated primarily with fixed edgewise
appliances.
These results indicate that there is a strong correlation between occlusal index scores at T1 and postretention stability. This information suggests that we
should use long-term retention for patients with more
severe initial malocclusions.
Absent from the final multivariate predictive formula are factors related to the posttreatment occlusion
(eg, PAR and ABO at T2). Although the posttreatment
PAR, II, and ABO scores were all lower for the stable
group, these differences were not large (less than 2
unweighted PAR points, less than 0.5 mm of irregularity, and less than 7 ABO points). This suggests that
patients from both groups were finished to comparable
levels, and that the posttreatment condition was not the
most influential factor in long-term stability. Although
some believe that the perfection of the occlusion
reduces the likelihood of subsequent relapse,18-20 the
analysis of the ABO scores at T2 shows a regression to
the mean, in which subjects with low T2 scores tend to
increase (deteriorate) at T3, and those with high T2
scores tend to decrease (improve) at T3.
Several studies contradict our results, finding that
finish quality had greater effects on stability. KahlNieke et al29 concluded that Class II or III molar
relationships that remained at the end of treatment were

Ormiston et al 573

associated with postretention changes of incisor alignment. The limitations of this study were discussed
previously. Birkeland et al15 found that PAR scores at
T2 were associated with long-term stability, but the
strength of this association was questionable (R .31).
The ABO was much more critical in scoring occlusions
at T2; this might explain the difference in conclusions
when the PAR and ABO indexes are used. Parkinson et
al30 found that patients with the greatest amount of
occlusal contact at T2 had the least amount of overbite
increase between T2 and T3, but they used some
individual measurements in their study. The use of a
single comprehensive occlusal index such as the PAR
or the ABO might have made comparisons to our result
more meaningful.
Several recent studies support our result. Otuyemi
and Jones21 found that PAR scores in well-treated
patients tended to deteriorate over time. Only 38% of
the patients had maintained their treatment results at T3
as measured by the PAR index. In relating T2 occlusal
discrepancies with T3 irregularity index changes, Weiland31 noted that occlusal slides (shifts) decreased from
T2 to T3, suggesting that some problems at T2 will
diminish in severity by T3. The scores obtained with
the ABO index reflected a similar pattern; the ABO
scores tended to regress toward the mean. Little et al16
also showed a similar tendency for changes in occlusal
relationships. Kashner24 (using the British weighting of
the PAR) and Fernandes23 found no correlation between level of treatment quality and long-term stability.
Our results suggest that finishing occlusal relationships to perfection might not ensure postretention
stability. If this is true, other factors might play more
important roles in stability, such as treatment and
retention methods, compliance, the tendency to return
to the pretreatment condition, growth, and other factors
that we might be unaware of. Additionally, there was
considerable individual variation in stability after
orthodontic treatment.
Our study had some limitations. Information about
treatment and retention was limited to what was available in the patient records. Various cephalometric
machines had been used, and quantifying magnification
between radiographs was problematic for linear measurements. The number of patients with restorations
generally increased over time; this might alter or mask
marginal ridge discrepancies and true alignment of
teeth. No attempt was made to control for this problem.
Information about supracrestal fiberotomy was not
available from most patient charts, but this procedure
was not commonly used when most of these patients
were treated. In addition, by selecting the most stable
and the most unstable adolescent subjects for the

574 Ormiston et al

sample, we chose 2 disparate groups. One might argue

that we should be most concerned with patients who are
likely to experience considerable relapse, because the
ability to predict these outcomes might influence our
treatment or retention decisions. This investigation is
helpful in that sense. However, a larger study including
subjects with the entire range of PAR outcomes would
result in a model that could be generalized to a broader
population.
CONCLUSIONS

1. In this sample, male sex and greater facial growth

were related, and both were associated with increased instability.
2. The initial severity of the malocclusion, as graded
by the PAR index and the II was correlated with
postretention instability. Patients with more severe
pretreatment index scores tended to be less stable.
3. High-quality treatment results, measured by ABO
scores, tended to deteriorate. Lower-quality treatment tended to improve. Thus, the differences in
the ABO scores at T2 were diminished by T3 .
We thank the University of Washington Orthodontic Alumni Association for supporting this project.

REFERENCES
1. Blake M, Bibby K. Retention and stability: a review of the
literature. Am J Orthod Dentofacial Orthop 1998;114:299-306.
2. Isaacson RJ, Zapfel RJ, Worms FW, Erdman AG. Effects of
rotational jaw growth on the occlusion and profile. Am J Orthod
1977;72:276-86.
3. Schudy GF. Posttreatment craniofacial growth: its implications
in orthodontic treatment. Am J Orthod 1974;65:39-57.
4. Siatkowski RE. Incisor uprighting: mechanism for late secondary
crowding in the anterior segments of the dental arches. Am J
Orthod 1974;66:398-410.
5. Sampson WJ. Current controversies in late incisor crowding.
Ann Acad Med Singapore 1995;24:129-37.
6. McReynolds DC, Little RM. Mandibular second premolar extractionpostretention evaluation of stability and relapse. Angle
Orthod 1991;61:133-44.
7. Driscoll-Gilliland J, Buschang PH, Behrents RG. An evaluation
of growth and stability in untreated and treated subjects. Am J
Orthod Dentofacial Orthop 2001;120:588-97.
8. Sinclair PM, Little RM. Dentofacial maturation of untreated
normals. Am J Orthod 1985;88:146-56.
9. Shields TE, Little RM, Chapko MK. Stability and relapse of
mandibular anterior alignment: a cephalometric appraisal of
first-premolar-extraction cases treated by traditional edgewise
orthodontics. Am J Orthod 1985;87:27-38.
10. Harris EH, Gardner RZ, Vaden JL. A longitudinal cephalometric
study of postorthodontic craniofacial changes. Am J Orthod
Dentofacial Orthop 1999;115:77-82.

American Journal of Orthodontics and Dentofacial Orthopedics

November 2005

11. Kim TW, Little RM. Postretention assessment of deep overbite

correction in Class II Division 2 malocclusion. Angle Orthod
1999;69:175-86.
12. Simons ME, Joondeph DR. Change in overbite: a ten-year
postretention study. Am J Orthod 1973;64:349-67.
13. Uhde MD, Sadowsky C, BeGole EA. Long-term stability of
dental relationships after orthodontic treatment. Angle Orthod
1983;53:240-52.
14. Berg R. Post-retention analysis of treatment problems and
failures in 264 consecutively treated cases. Eur J Orthod 1979;
1:55-68.
15. Birkeland K, Furevik J, Boe OE, Wisth PJ. Evaluation of
treatment and post-treatment changes by the PAR Index. Eur
J Orthod 1997;19:279-88.
16. Little RM, Wallen TR, Riedel RA. Stability and relapse of
mandibular anterior alignment-first premolar extraction cases
treated by traditional edgewise orthodontics. Am J Orthod
1981;80:349-65.
17. Fidler BC, rtun J, Joondeph DR, Little RM. Long-term stability
of Angle Class II, Division 1 malocclusions with successful
occlusal results at end of active treatment. Am J Orthod Dentofacial Orthop 1995;107:276-85.
18. Andrews LF. The six keys to normal occlusion. Am J Orthod
1972;62:296-309.
19. Pancherz H. The effects, limitations, and long-term dentofacial
adaptations to treatment with the Herbst appliance. Semin Orthod
1997;3:232-43.
20. Solow B. The dentoalveolar compensatory mechanism: background and clinical implications. Br J Orthod 1980;7:145-61.
21. Otuyemi OD, Jones SP. Long-term evaluation of treated Class II
Division 1 malocclusions utilizing the PAR index. Br J Orthod
1995;22:171-8.
22. Little RM, Riedel RA, rtun J. An evaluation of changes in
mandibular anterior alignment from 10 to 20 years postretention.
Am J Orthod Dentofacial Orthop 1988;93:423-8.
23. Fernandes A. The effect of quality of treatment results on long
term stability [thesis]. Los Angeles: University of Southern
California; 2000.
24. Kashner JE. Does higher quality of treatment translate into a
higher quality long-term result? [thesis]. Seattle: University of
Washington; 1999.
25. Richmond S, Shaw WC, Roberts CT, Andrews M. The PAR
index (peer assessment rating): methods to determine outcome of
orthodontic treatment in terms of improvement and standards.
Eur J Orthod 1992;14:180-7.
26. Harvold EP, Vargervik K. Morphogenetic response to activator
treatment. Am J Orthod 1971;60:478-90.
27. Sinclair PM, Little RM. Maturation of untreated normal occlusions. Am J Orthod 1983;83:114-23.
28. Vaden JL, Harris EF, Gardner RL. Relapse revisited. Am J
Orthod Dentofacial Orthop 1997;111:543-53.
29. Kahl-Nieke B, Fischbach H, Schwarze CW. Post-retention
crowding and incisor irregularity: a long-term follow-up evaluation of stability and relapse. Br J Orthod 1995;22:249-57.
30. Parkinson CE, Buschang PH, Behrents RG, Throckmorton GS,
English JD. A new method of evaluating posterior occlusion and
its relation to posttreatment occlusal changes. Am J Orthod
Dentofacial Orthop 2001;120:503-12.
31. Weiland FJ. The role of occlusal discrepancies in the long-term
stability of the mandibular arch. Eur J Orthod 1994;16:521-9.