Upload
fiorella-loli-pinto
View
215
Download
0
Embed Size (px)
Citation preview
8/14/2019 1-s2.0-S0889540609009329-main
1/14
ONLINE ONLY
Correlations between cephalometric andphotographic measurements of facial
attractiveness in Chinese and US patients afterorthodontic treatment
Hee Soo Oh,a Edward L. Korn,b Xiaoyun Zhang,c Yan Liu,c Tianmin Xu,d Robert Boyd,e and Sheldon Baumrindf
San Francisco, Calif, Bethesda, Md, and Beijing, China
Introduction:Orthodontists rely on esthetic judgments from facial photographs. Concordance between es-
timates of facial attractiveness made from lateral cephalograms and those made from clinical photographs
has not been determined. We conducted a preliminary examination to correlate clinicians rankings of facial
attractiveness from standardized end-of-treatment facial photographs (Photo Attractiveness Rank) with ceph-
alometric measurements of facial attractiveness made for the same subjects at the same time. Methods:
Forty-five Chinese and US orthodontic clinicians ranked end-of-treatment photographs of separate samples
of 45 US and 48 Chinese adolescent patients for facial attractiveness. Separately for each sample, the pho-
tographic rankings were correlated with the values of 21 conventional hard- and soft-tissue measures from
lateral cephalograms taken at the same visits as the photographs. Results: Among US patients, higher
rank for facial attractiveness on the photographs was strongly associated with higher values for profile angle,
chin prominence, lower lip prominence, and Z-angle, and also with lower values for angle of convexity, H-an-
gle, and ANB. Among Chinese patients, higher rank for facial attractiveness on the photographs was strongly
associated with higher values for Z-angle and chin prominence, and also with lower values for angle of con-
vexity, H-angle, B-line to upper lip, and mandibular plane angle. Chinese patients whose %lower face height
values approximated the ethnic ideal (54%) tended to rank higher for facial attractiveness than patients with
either higher or lower values for %lower face height. The absolute values of the correlations for the 7 US mea-
sures noted above ranged from 0.41 to 0.59; those of the 7 Chinese measures ranged from 0.39 to 0.49.TheP
value of the least statistically significant of these 14 correlations was 0.006, unadjusted for multiple compar-
isons. On the other hand, many cephalometric measures believed by clinicians to be indicators of facial attrac-tiveness failed to correlate with facial attractiveness rank for either ethnicity at even the P\0.05 level, including
SN-pogonion angle, lower incisor to mandibular plane angle, and Wits appraisal. Conclusions: In general,
there was less association than expected or desired between objective measurements on the lateral cephalo-
grams and clinicians rankings of facial attractiveness on sets of clinical photographs. (Am J Orthod
Dentofacial Orthop 2009;136:762.e1-762.e14)
Much of the orthodontic literature since Downs
classic 1948 article has focused on the analy-
sis of the lateral cephalograms.1 Combina-
tions of cephalometric measures have been grouped
into many cephalometric analyses (eg, those of
Downs,1,2 Steiner,3 Tweed,4,5 Jarabak andFizzell,6 Sas-
souni,7 Bjork,8 Ricketts,9,10 McNamara,11 Arnett and
Bergman,12 and Arnett et al13). Findings from each of
these analyses are used by orthodontists in diagnosis,
treatment planning, and outcomes research. Almost all
aAssistant professor, Department of Orthodontics, Arthur A. Dugoni School of
Dentistry, University of the Pacific, San Francisco, Calif.bMathematical statistician, Biometric Research Branch, National Cancer Insti-
tute, NIH, Bethesda, Md.cAssistant professor, Department of Orthodontics, PekingUniversity School and
Hospital of Stomatology, Beijing, China.dProfessor and chair, Department of Orthodontics, Peking University School
and Hospital of Stomatology, Beijing, China.eProfessor and chair, Department of Orthodontics, Arthur A. Dugoni School of
Dentistry, University of the Pacific, San Francisco, Calif.fProfessor, Department of Orthodontics, and Director, Craniofacial Research
Instrumentation Laboratory, Arthur A. Dugoni School of Dentistry, University
of the Pacific, San Francisco, Calif.
The authors report no commercial, proprietary, or financial interest in the prod-
ucts or companies described in this article.
Supported by a grant of Orthodontic Faculty Development Fellowship Award
from the American Association of Orthodontists Foundation and in part by
NIHNIDR Grants DE07332 and DE08713.
Reprint requests to: Sheldon Baumrind, 1525 Walnut St, Berkeley, CA 94709-
1512; e-mail,[email protected].
Submitted, January 2009; revised and accepted, April 2009.
0889-5406/$36.00
Copyright 2009 by the American Association of Orthodontists.
doi:10.1016/j.ajodo.2009.04.020
762.e1
mailto:[email protected]:[email protected]8/14/2019 1-s2.0-S0889540609009329-main
2/14
these analyses contain some measurements that are be-
lieved to quantify facial attractiveness. In recent
years, orthodontists have also come to rely heavily on es-
thetic judgments from semistandardized facial photo-
graphs. These photographs have now becomea component of the orthodontists routine records set
for diagnosis, treatment planning, and outcome analysis.
An examination of the concordance between esti-
mates of facial attractiveness made from lateral cepha-
lograms and those made from clinical photographs is
therefore in order. Some previous studies have probed
this relationship, but the available findings are far
from conclusive.14-16 In our study, we conducted a pre-
liminary examination of this clinically important ques-
tion by correlating clinicians rankings of facial
attractiveness from standardized end-of-treatment fa-
cial photographs (Photo Attractiveness Rank) with
cephalometric measurements of facial attractiveness
made for the same subjects at the same time. Our gen-
eral hypothesis was that thevalues for the cephalometricmeasures would correlate highly with Photo Attractive-
ness Rank. In testing this hypothesis, we sought explicit
answers to the following specific questions.
1. To what extent do conventional hard- and soft-tis-
sue cephalometric measures generally considered
to be related to facial attractiveness correlate with
Photo Attractiveness Rank?
2. To what extent do the data support the idea that the
patients with the highest Photo AttractivenessRanks have cephalometric values close to the ceph-
alometric ideals, whereas patients with cephalo-metric values either higher or lower than the
cephalometric ideals tend to be judged less at-
tractive? (We asked this question because it reflects
the common expectation among orthodontists that,
for many cephalometric measures, the highest rank
for facial attractiveness on photographs would be
closely associated with the ideal value of the ceph-
alometric measures, whereas both higher and lower
values for the cephalometric measures would tend
to be associated with lower Photo Attractiveness
Ranks. To the extent that this condition holds for
any particular cephalometric measure, its presence
will not be detected by conventional linear correla-
tional methods unless the data are transformed.
This problem and its solution will be discussed in
Material and methods.)
3. To what extent does adding together information
from several different cephalometric measures of
facial attractiveness improve our ability to account
for observed differences in Photo Attractiveness
Rank? (We asked this question to test the assump-
tion that the merging of information from different
cephalometric measures improves our ability to
predict Photo Attractiveness Rank.)
Answers to these questions were sought by testing
them in hypothesis form against data from samples oflateral cephalograms and facial photographs comprising
a total of 93 treated Chinese andUS orthodontic patients.
MATERIAL AND METHODS
The data evaluated in this study are from 2 samples
of orthodontic patients of different ethnicities treated by
experienced orthodontists in the United States and
China. They were randomly selected from the records
of patients who had completed treatment at the graduate
orthodontic clinic of the Department of Orthodontics at
Peking University, Beijing, China, and the private prac-tice of the director of the graduate orthodontic clinic of
the New Jersey Dental School, Newark.
At each institution, a stratified random sample of 48
subjects was acquired, comprising 4 groups of 12 pa-
tients each. In each group of 12 patients, 3 were Class
I extraction, 3 Class I nonextraction, 3 Class II extrac-
tion, and 3 Class II nonextraction. The sampling process
was the same at both institutions and is illustrated sche-
matically inFigure 1. Three of the original 48 US sub-
jects were excluded from this study for poor image
quality, yielding a final sample of 45. In addition, sev-
eral of the remaining US subjects hae missing valuesfor some cephalometric measures, but the sample size
for no cephalometric measure was less than 41.
The US sample consisted of late-adolescent whitepatients living in northern New Jersey. The Chinese
sample consisted of late-adolescent patients of Chinese
ethnicity, residing in the Beijing area. Cephalometric
and photographic data were complete for all subjects
in the Chinese sample. Further details of the sampling
process have been supplied in previous articles.17,18
Among the end-of-treatment materials available for
each subject were conventional lateral cephalogramsand standardized triplet facial photographs taken at
the same end-of-treatment visit (Fig 2). The photo-
graphs of the US sample were taken with the patient po-
sitioned in the cephalostat according to the original
standard of the Second Roentgenographic Workshop.19
The triplet photographs from both samples had been
ranked for facial attractiveness by each of 45 orthodon-
tically trained judges (25Chinese, 20 US) using previ-
ously described methods.17,18 Also available for each
patient were the database-stored coordinates of 35
hard- and soft-tissue landmarks located on the end-of-
treatment lateral cephalogram for each subject. Each
762.e2 Oh et al American Journal of Orthodontics and Dentofacial OrthopedicsDecember 2009
8/14/2019 1-s2.0-S0889540609009329-main
3/14
cephalogram had been traced independently by 2 or
more judges, and the average values had been recorded
in a numerical database from which the measures of
almost all standard cephalometric analyses could becalculated by computer operations.
From the coordinate values of a subset of the hard-and
soft-tissue landmarks, 9 cephalometric measures whose
values are conventionally considered to be strong indica-
tors of facial attractiveness were selected by the principal
author (H.O.) for analysis. Later, an informal survey con-
ducted among faculty orthodontists at the Universityof the
Pacific in San Francisco identified 12 additional estheti-
cally related cephalometric variables, which were added
to the analysis. To reduce the complexity of data presenta-tion, the results of the 2 analyses have been combined.
Operational definitions for the 21 measures are
listed below. Among them are 10 hard-tissue measures
(designated [H]), 3 orientation-free soft-tissue measures
(designated [S]), and 8 soft-tissue measures oriented to
the hard-tissue-defined Frankfort line (designated [SF]).
Eleven measures are angular; 9 are linear, and 1 is a ra-
tio. The 9 measures examined in the first pass are iden-
tified by asterisks.
Angular measures (in degrees)
1. ANB angle*: the included anglebetween Point A,
nasion, and Point B (after Riedel20) [H].
2. Angle of convexity: theincluded angle between na-
sion, Point A, and pogonion (after Downs2) [H].
3. Facial plane angle*: the acute angle formed by the
intersection of the line nasion-pogonionwith theFrankfort horizontal line (after Holdaway21) [H].
4. H-angle: the included angle between soft-tissue
nasion, soft-tissue pogonion, and the most anterior
point on the upper lip (after Holdaway22) [S].
5. Lower incisor to A-pogonion angle: the acute
angle formed by the intersection between the
line Point A-pogonion and the long axis of the
lower incisor (after Ricketts23) [H].
6. Lower incisor to mandibular plane angle*: the an-
gle formed by the intersection of the long axis of
the lower central incisor andthe lower border of
the mandible (after Tweed4,5) [H].
7. Mandibular plane angle*: the acute angle formed
by the intersection of the line gonion-menton with
the Frankfort horizontal plane (after Downs2) [H].
8. Nasolabial angle: the included angle between
lower nose tip, soft-tissue subnasale, and the ver-
milion border of the upper lip (after Burstone24)
[S].
9. Profile angle*: the included angle between gla-bella (defined in this study as the most anterior
point on the soft tissue of the forehead), soft-tissue
subnasale, and soft-tissue pogonion, defined in
Fig 1. Schematic showing the procedures for drawing the equivalent Chinese and US samples. Step
1: at each venue, patients who received treatment during a specified time period were identified. Step
2: each patient wasassigned a random number; all subsequent procedures were conducted with the
charts sorted in random order to ensure that the sample was representative of the population (ie,
practice) from which it was drawn. Step 3: proceeding in random order, all charts with complete re-
cords were identified and duplicated for further studies (for the purposes of the general project of
which this study is a part, a complete record was considered to be one in which a lateral cephalo-
gram, study casts, full-mouth intraoral or panoramic x-rays, and a facial photographic triplet were
available at the beginning and end of full-bonded orthodontic treatment). Step 4: parallel stratified
subsets of 48 Chinese subjects and 45 US subjects with complete records were selected; the subset
from each institution comprised the first 48 randomly ordered charts from that venue that satisfied the
Angle Class and extraction and nonextraction criteria for this study, except that 3 subjects were lost
from the US group because their images were technically unsatisfactory.
American Journal of Orthodontics and Dentofacial Orthopedics Oh et al 762.e3Volume136,Number6
8/14/2019 1-s2.0-S0889540609009329-main
4/14
this study as the most anterior point onthe soft-tis-
sue surface of the chin (after Burstone25) [S].
10. SN-pogonion angle*: the included angle between
the landmarks sella-nasion and pogonion (after
Bell et al26) [H].
11. Z-angle: the angle formed by the intersection be-
tween the Frankfort horizontal line and the line
from soft-tissue pogonion to the anterior-most
point on the lower lip (after Merrifield27) [SF].
Linear measures (in millimeters)
12. B-line to lower lip: the distance from the anterior-
most point on the lower lip and the line from soft-
tissue subnasale to soft-tissue pogonion measured
parallel to the Frankfort plane (after Burstone25)
[SF].
13. B-line to upper lip: the distance from the anterior-
most point on the upper lip and the line from soft-
tissue subnasale to soft-tissue pogonion measured
parallel to the Frankfort plane (after Burstone25)
[SF].
14. Chin prominence: the distance from soft-tissue
pogonion to a line perpendicular to the Frankfort
plane and passing through soft-tissue subnasale
(after Arnett et al13) [SF].
15. E-line to lower lip*: the distance from the ante-
rior-most point on the lower lip to the line from
the nose tip to soft-tissue pogonion measured
parallel to the Frankfort plane (after Ricketts23)
[SF].
16. Lower incisor to A-pogonion distance: the dis-
tance between the tip of the lower incisor and
a line from Point A to pogonion measured parallel
to the Frankfort plane (after Ricketts23) [H].
Fig 2. Representative photographic triplets from the US (upper) and Chinese (lower) samples.
762.e4 Oh et al American Journal of Orthodontics and Dentofacial OrthopedicsDecember 2009
8/14/2019 1-s2.0-S0889540609009329-main
5/14
17. Lower lip prominence: the distance from the ante-
rior-most point of the lower lip to a line perpendic-
ular to the Frankfort plane and passing through
soft tissue subnasale (after Arnett et al13) [SF].
18. Nose prominence: the distance from the nose tip to
a line perpendicular to the Frankfort plane and
passing through soft-tissue subnasale (after Arnett
et al13) [SF].
19. Upper lip prominence: the distance from the ante-
rior-most point of the upper lip to a line perpendic-
ular to the Frankfort plane and passing through
soft-tissue subnasale (after Arnett et al13) [SF].20. Wits appraisal*: the distance between the intersec-
tion of a perpendicular line to occlusal plane from
Point A and the intersection of a perpendicular
line to the occlusal plane from Point B (after
Jacobson28) [H].
Ratio
21. %Lower face height*: lower face height times 100
divided by total face height (after Farkas29) [H].
The locations of the anatomical landmarks required
to compute the 21 measures are shown inFigure 3.
Statistical analysis
Since this study was essentially an exploratory in-
vestigation of a new area of interest, we allowed our-
selves to consider more variables than would be
appropriate in a definitive hypothesis test. Rather than
perform a formal multiple comparisons adjustment,
we chose to focus our consideration on measures whose
Pvalues were less than 0.01.For answers to the first question, Pearson correla-
tions (r) were used to calculate the strength of associa-tion between Photo Attractiveness Ranks of the
patients and their scores for each of 21 cephalometric
measures.
Separate tests were performed for the patients of
each ethnicity. However, each patients Photo Attrac-
tiveness Rank was the single mean value for all 45 US
and Chinese judges pooled. The decision to pool the
values of the Chinese and US judges was based on the
previous finding that the correlation between the Photo
Fig 3. Locations of the landmarks used in computing the 21 cephalometric measures. Thedashed
lines show the Frankfort horizontal and a line perpendicular to it passing through subnasale.
American Journal of Orthodontics and Dentofacial Orthopedics Oh et al 762.e5Volume136,Number6
8/14/2019 1-s2.0-S0889540609009329-main
6/14
Attractiveness Ranks of the Chinese and US judges was
strong (r 50.92 and P \0.0001 for USpatients; r 50.86 and P \0.0001 for Chinese patients).18
In examining associations between variables, we
have chosen to report Pearson rho values rather than
Spearman rho values that are frequently used for ranked
data. We reasoned that, since Photo Attractiveness Rank
for each patient in this study is the mean of 45 individual
values, they would be expected to be approximately nor-
mally distributed by the central limit theorem. In addi-tion, the cephalometric variables did not appear to
have grossly nonnormal distributions. Therefore, thePearson correlation would be expected to have good
properties for detecting linear associations between
the Photo Attractiveness Rank and the cephalometric
variables. In theory, the Spearman rho might be more
sensitive to nonlinear but monotone associations, al-
though with the small sample sizes this is unclear. Nei-
ther correlation would be sensitive to nonmonotone
associations; this is why we used the absolute-value
transformations to address question 2.
The advantage of using the Pearson correlation over
the Spearman was that its square (r2) can be used as
a measure of explained variance, thus facilitating
a straightforward assessment of increased association
when combinations of cephalometric variables were
considered in question 3.
Question 3 was investigated by using the technique
of stepwise linear regression. A P value less than 0.01
was required for a variable to enter the model for each
ethnicity, and a P value greater than 0.05 was requiredto leave the model. Although a number of alternative
methods of regression analysis were available, we con-
sidered stepwise regression a reasonable approach for
assessing the association with combinations of cephalo-
metric variables, given the relatively small sample sizes
and the exploratory nature of the study.
RESULTS
Table I lists the means and standard deviations for
all 21 cephalometric measures for the US and Chinese
Table I. Means and standard deviations by ethnicity for this sample and means of best available comparable studies
US cohort Chinese cohort
Present sample Best avail able norm Present sampl e Best available norm
Cephalometric measures Mean SD Mean SD Ref Mean SD Mean SD Ref
Angular measures ()
ANB 2.7 1.9 2.0 1.8 20 4.0 2.3 2.5 1.5 30
Angle of convexity 2.4 5.3 0.0 5.1 2 7.9 5.7 5.5 2.9 31
Facial plane angle 83.5 3.4 85.0 3.0 21 83.1 3.5 87.7 3.0 30
H-angle 13.5 3.7 10.0 3.0 22 18.0 3.7 13.6 3.6 30
Lower incisor to A-pogonion line 25.7 5.2 22.0 4.0 23 27.9 6.5 26.9 5.1 2
Lower incisor to mandibular plane 95.6 7.0 90.0 5.8 4, 5 99.0 9.6 94.3 5.5 30
Mandibular plane angle 27.7 6.6 25.0 4.5 4, 5 30.1 7.1 27.2 4.7 30
Nasolabial angle 127.3 8.2 102.0 8.0 24 97.1 9.6 84.0 80-110 31
Profile angle 165.6 5.9 168.7 4.1 25 166.6 5.1 168.6 3.7 31
SN-pogonion angle 77.0 3.7 79.0 3.0 26 76.7 3.7 79.5 3.3 30
Z-angle 71.2 6.9 80.0 9.0 27 66.3 6.2 71.0 32
Linear measures (mm)
B-line to lower lip 2.6 2.5 3.0 1.0 25 5.9 2.3 6.5 1.6 30B-line to upper lip 3.1 1.8 2.0 1.0 25 6.1 1.7 4.9 1.7 30
Chin prominence (F) 12.2 5.4 2.6 1.9 13 11.8 5.2 6.7 3.3 31
(M) 15.6 8.6 3.5 1.9 16.2 4.6 8.5 3.4
E-line to lower lip 2.9 3.2 2.0 2.0 23 0.9 2.4 1.4 1.9 30
Lower incisor to A-pogonion line 2.4 1.6 1.0 2.0 23 4.4 2.0 4.3 1.8 31
Lower lip prominence (F) 4.5 3.1 1.9 1.4 13 0.1 2.8 3.8 2.3 31
(M) 4.3 4.2 1.0 2.2 3.4 3.2 2.6 1.6
Nose prominence 17.1 2.4 16.0 1.4 13 14.9 2.1 13.1 2.0 31
Upper lip prominence (F) 0.5 2.1 3.7 1.2 13 1.8 2.1 6.1 2.2 31
(M) 0.0 2.6 3.3 1.7 0.0 2.3 5.5 2.4
Wits appraisal (F) 2.3 4.0 0.1 1.8 28 0.9 3.1 3.9 2.1 31
(M) 3.2 4.3 1.2 1.9 0.1 2.1 2.6 1.2
Ratio (%)
%Lower face height 53.7 2.3 55.5 1.0 29 53.7 1.9 54.0 1.0 31
Ref, Reference; F, female;M, male.
762.e6 Oh et al American Journal of Orthodontics and Dentofacial OrthopedicsDecember 2009
8/14/2019 1-s2.0-S0889540609009329-main
7/14
samples. For use in later comparisons, it also contains
information on the ideal value for each measure for
each ethnicity.2,4,5,2032
The findings of the study will be presented in terms
of the 3 questions raised in the introduction.
Question 1 was to what extent do conventional hard-
and soft-tissue cephalometric measures generally con-
sidered to be related to facial attractiveness correlate
with clinicians rankings of facial photographs for facial
attractiveness?
To answer this question,Table II lists the correlation
with the Photo Attractiveness Rank of each of the 21cephalometric measures for each ethnicity. Within eth-
nicity, the values are listed in descending order of abso-
lute magnitude.
Seven cephalometric measures for the US sample
and 6 cephalometric measures for the Chinese sample
correlated with the judges rankings of facial photo-
graphs withPvalues less than 0.01. The absolute values
of the correlations for the 7 US measures ranged from
0.41 to 0.59. The corresponding range for the 6 Chinese
measures was 0.39 to 0.49. For each ethnicity, both lin-
ear and angular measures and both hard- and soft-tissue
measures were found among the most highly significant
measures. Four of the most highly significant measures
were the same in the Chinese and US patient samples:
angle of convexity, chin prominence, Z-angle, and
H-angle.
In the US sample, profile angle, chin prominence,lower lip prominence, and Z-angle correlated positively
with Photo Attractiveness Rank. As the value of each of
these measures increased, the patient tended to be
judged more attractive. On the other hand, angle of con-
vexity, H-angle, and ANB angle correlated negatively
with the Photo Attractiveness Rank. As the value ofeach of these measures increased, the patient tended
to be judged less attractive.
In the Chinese sample, Z-angle and chin promi-
nence correlated positively with Photo Attractiveness
Rank. As the value of each of these cephalometric mea-
sures increased, the patient tended to be judged more at-
tractive. On the other hand, angle of convexity, H-angle,
B-line to upper lip, and mandibular plane angle corre-
lated negatively with Photo Attractiveness Rank; as
the value of each of these cephalometric measures in-
creased, the patient tended to be judged less attractive.
Table II. Correlations between Photo Attractiveness Rank and 21 raw cephalometric measures
US patients (n 5 45) Chinese patients (n 5 48)
Order Cephalometric measure
Raw cephalometric value
Order Cephalometric measure
Raw cephalometric value
r P r P
1 Profile angle () 0.59 \0.0001 1 Z-angle () 0.49 0.0004
2 Angle of convexity () 0.51 0.0003 2 Angle of convexity () 0.45 0.002
3 Chin prominence (mm) 0.50 0.001 3 Chin prominence (mm) 0.42 0.003
4 H-angle () 0.48 0.001 4 H-angle () 0.40 0.005
5 ANB () 0.44 0.003 5 B-line to upper lip (mm) 0.40 0.005
6 Lower lip prominence (mm) 0.41 0.005 6 Mandibular plane angle () 0.39 0.006
7 Z-angle () 0.42 0.006 7 ANB () 0.35 0.014
8 Mandibular plane angle () 0.38 0.011 8 Facial plane angle () 0.34 0.02
9 Upper lip prominence () 0.32 0.03 9 E-line to lower lip (mm) 0.34 0.02
10 Lower incisor to A-pogonion line
(mm)
0.29 0.05 10 Lower incisor to A-pogonion line () 0.33 0.02
11 E-line to lower lip (mm) 0.29 0.06 11 Profile angle () 0.32 0.03
12 Nose prominence (mm) 0.25 0.10 12 B-line to lower lip (mm) 0.31 0.03
13 SN-pogonion angle (
) 0.24 0.11 13 Lower incisor to mandibular plane()
0.25 0.09
14 Nasolabial angle () 0.25 0.13 14 Nose prominence (mm) 0.20 0.18
15 Facial plane angle () 0.23 0.14 15 Lower lip prominence (mm) 0.19 0.20
16 Wits appraisal (mm) 0.21 0.16 16 Upper lip prominence (mm) 0.10 0.50
17 B-line to lower lip (mm) 0.19 0.22 17 SN-pogonion angle () 0.10 0.52
18 B-line to upper lip (mm) 0.12 0.43 18 %Lower face height (%) 0.09 0.55
19 Lower incisor to mandibular plane
()
0.07 0.65 19 Nasolabial angle () 0.06 0.69
20 Lower incisor to A-pogonion line () 0.05 0.73 20 Wits appraisal (mm) 0.02 0.87
21 %Lower face height (%) 0.01 0.94 21 Lower incisor to A-pogonion line
(mm)
0.001 0.99
Negative values for r indicate cephalometric measures that decrease as facial attractiveness increases.
American Journal of Orthodontics and Dentofacial Orthopedics Oh et al 762.e7Volume136,Number6
8/14/2019 1-s2.0-S0889540609009329-main
8/14
Scatterplots for each of the highly statistically sig-
nificant measures in Table II are given in Figure 4.
They show the distributions of individual case values
for the associations between Photo Attractiveness
Rank on the y-axis and each of the cephalometric mea-sures on the x-axis. Values for the least attractive pa-
tients are at the bottom of each plot, and values for
the most attractive patients are at the top. Within ethnic-
ity, the y-value for each patient is the same in all plots.
On each plot, the ideal value of the cephalometric
measure is represented by the vertical red line. It can
be seen that, for almost all these measures, the ideal
value falls within the distribution, but it is rarely near
the distributions center. For the majority of cephalo-
metric variables, the values for the less attractive pa-
tients tend to be farther from the ideal than the
values for the more attractive patients.
The patterns of the scatterplots for each of the 4 vari-
ables that met our high significance criterion for both
ethnicities (angle of convexity, chin prominence, H-an-
gle, and Z-angle) were quite similar for both ethnicities.
With regard to the goal of clinical prediction, it was
noted that, even though when the correlations are highly
statistically significant, these real-world distributions
showed considerable variability.
We also thought it important that many cephalomet-
ric measures generally considered to be strongly associ-
ated with facial attractiveness failed to correlate
significantly with the judges rankings of facial attrac-
tiveness on photographs. Among these were lower inci-sor to mandibular plane angle, %lower face height, Wits
appraisal, nose prominence, and all other measures in
Table I whose probability of occurring by chance was
greater than 0.05. At this point in the study, it was not
possible to be sure whether the values for these variables
actually had little association with Photo Attractiveness
Rank or whether they had a curvilinear association that
was undetectable by linear correlation unless the data
were transformed. Hence, the relevance of Question 2.
Question 2 was to what extent do the data support
the idea that the patients with the highest Photo Attrac-
tiveness Ranks have cephalometric values close to thecephalometric ideals, whereas patients with cephalo-
metric values either higher or lower than the cephalo-
metric ideals tend to be judged less attractive?
If we correlate Photo Attractiveness Rank with
a cephalometric measure whose facial attractiveness is
highest at its ideal value but lower both above and
below the ideal value (a condition we expected to
encounter for a number of cephalometric measures),
the regression line would tend to approximate a para-
bolic curve rather than the straight line that results
when 2 variables maintain the same relationship to
each other throughout their range.
This is a problem because the Pearson correlation
method as usually used is almost completely insensitive
to this type of nonlinear distribution. To protect our-selves from failing to detect associations of this type be-
tween Photo Attractiveness Rank and one or more of the
cephalometric measures under examination, a new
value was calculated equal to the absolute distance (in
degrees or millimeters) between each patients raw
value for each measure and the ethnic ideal for that
measure. The transformed cephalometric values for
each cephalometric measure are then amenable to con-
ventional linear correlation with the patients Photo
Attractiveness Ranks.
The ideal value for each measure for each ethnicity
from Table I was used to transform the raw value of each
subject by rescaling it in terms of its absolute distance
from the ideal.Table III lists the transformed correla-
tions and P values for the relationships between Photo
Attractiveness Rank and each of the 21 cephalometric
measures for both the Chinese and the US samples.
Also listed is the increment or decrement in the strength
of each correlation as compared withTable II. As can be
seen in the table, only 1 measure that did not satisfy our
P\0.01 criterion for statistical significance inTable II
achieved such significance after transformation. This
was the transformed value of %lower face height in
the Chinese sample. For this cephalometric measure,
the change in r value was dramatic, moving from r 50.09 to r 5 0.44, and from the 18th position among
the 21 Chinese cephalometric variables to first. The ac-
tual distributions of the raw and transformed values for
this correlation are shown inFigure 5. This implies that
%lower face height values that were larger or smaller
than the Chinese ideal of 54% tended to be equally
unattractive.
No other cephalometric measure for either ethnicity
whose correlation with Photo Attractiveness Rank did
not satisfy the P \0.01 criterion before transformation
increased sufficiently to satisfy it after transformation.
B-line to lower lip and B-line to upper lip in the US sam-ple did show substantial improvements in r value after
transformation but still did not reach our criterion for
statistical significance.
In general, the transformations had less effect than
we had expected. Three of the 4 cephalometric mea-
sures that had correlated most highly with Photo Attrac-
tiveness Rank in both the Chinese and US samples in
Table I(chin prominence, angle of convexity, and Z-an-
gle) remained highly significant after transformation,
with only very small changes in r value. For example,
762.e8 Oh et al American Journal of Orthodontics and Dentofacial OrthopedicsDecember 2009
8/14/2019 1-s2.0-S0889540609009329-main
9/14
American Journal of Orthodontics and Dentofacial Orthopedics Oh et al 762.e9Volume136,Number6
8/14/2019 1-s2.0-S0889540609009329-main
10/14
the transformation increased thevalue of profile angle in
the US sample from 0.59 to 0.61.
It is theoretically possible that a curvilinear relation-
ship might have been missed by our transformation if
the ideal value for that cephalometric measure werefar from the center of the distribution of sampled values.
To control for this possibility, we examined the plots of
all cephalometric measures vs Photo Attractiveness
Rank. (All plots can be viewed on the Internet at
www.cril.org.) Only for lower lip prominence and naso-
labial angle, both for the US sample, were there any in-
dications that this might be true. To explore this
possibility further, we transformed the individual case
values of these 2 measures by computing their absolute
differences from their sample means rather than from
the ideal for the cephalometric measure. The result-
ing correlations were 0.25 (P 50.098) for lower lip
prominence and 0.39 (P 5 0.015) for nasolabial angle.
The value for lower lip prominence represents a decre-
ment with respect to the value inTable II. The finding
for nasolabial angle is more interesting. In the past,
our laboratory found the reliability of this measure to
be low because its landmarks are located near each other
and are ambiguous. The fact that the transformed value
nevertheless came so close to meeting our criterion for
statistical significance makes it a reasonable target for
further investigation.
Question 3 was are the associations made stronger
when combinations of cephalometric variables are com-
pared with the rankings from facial photographs?In clinical orthodontics, practitioners frequently ag-
gregate data from several measures. Hence, we asked in
our third question whether combinations among the sig-
nificant variables in our study can more fully account
for the observed variability in Photo Attractiveness
Rank.
To provide answers to this question, separate step-
wise regression analyses were performed for the US
and the Chinese samples. The dependent variable in
each test was Photo Attractiveness Rank. The indepen-
dent variables in each test were the most highly statisti-
cally significant raw and transformed cephalometricvariables for that ethnicity. We asked whether any com-
bination of these variables yielded a substantial im-
provement in accounting for attractiveness in either
ethnicity. For the US sample, the transformed value of
profile angle was the only variable that entered into
the stepwise regression model. Its r2 value was 0.37, im-
plying that slightly more than 2/5 of the variability in
Photo Attractiveness Rank could be accounted for by
the set of cephalometric variables.
For the Chinese sample, Z-angle entered the modelfirst, followed by the transformed value of %lower face
height. The cumulative r2 value for the 2 variables con-
sidered together was 0.41, implying that here too
slightly more than 2/5 of the variability in Photo Attrac-
tiveness Rank could be accounted for by the pair of
cephalometric variables. (The r2 value for Z-angle con-
sidered alone was 0.24.)
It can be seen that adding together information from
several raw and transformed cephalometric variables
did not greatly improve our ability to explain the clini-
cians rankings of facial photographs for attractiveness.
Although it is true that our sample sizes were small for
conducting multiple regression analyses of this sort, our
results do strongly imply the need for caution in treating
findings from different but related cephalometricmeasures as if information from them can be treated
additively.
DISCUSSION
Lateral cephalograms and facial photographs have
complementary roles in the evaluation of facial attrac-
tiveness by orthodontists. Photographs show the surface
structures of the face in considerable detail, and x-ray
images allow us to understand the relationship between
those surface structures and the skeletal and dental ar-mature that supports them. Certainly, photographs are
much closer to the natural state of the subject than are
lateral cephalograms. However, techniques for quantita-
tive measurement of facial photographs and standardi-
zation of photographic orientation are much less well
advanced in orthodontics.33-38 Ranking methods such
as the one that we used are useful for the conduct of
studies such as this one. But the numerical rank of
each of our subjects was completely dependent on the
judges evaluations of the other patients in our particular
sample. For that reason, it cannot be compared quantita-
tively to the rank of a photograph from any other sam-ple. Clearly, the development of reliable, clinically
useful quantitative and objective measures of facial at-
tractiveness on photographs would be desirable, but
such measures have yet to be developed. Now, only rel-
atively vague sets of rules and principles have been
Fig 4. Distributions of the actual relationship between Photo Attractiveness Rank and A, the 7 US
and B, 6 Chinese cephalometric measures for which highly significant associations are reported in
Table II. The purportedly ideal value for each measure from Table Iis indicated by a vertical red
line. Even though these associations are highly statistically significant, these real-world distributions
are much more scattered than the perfect schematic distributions.
:
762.e10 Oh et al American Journal of Orthodontics and Dentofacial OrthopedicsDecember 2009
http://www.cril.org/http://www.cril.org/8/14/2019 1-s2.0-S0889540609009329-main
11/14
established, mostlyimported from the fields of art (eg,
divine proportions,39-42
and vertical thirds and horizon-tal fifths43-46) and psychology (eg, averageness, 47-50
youthfulness,51,52 and symmetry53,54). Those measureson photographs that have been thus far developed in or-
thodontics (eg, the smile line55-57) have been subjected
to limited critical testing. And when tests have been
done, the results have generally not been statistically
significant.58,59
Lateral cephalograms, on the other hand, are fairly
well understood by the members of the orthodontic
community. Although quantification of facial esthetics
is certainly not the main use of cephalograms in ortho-
dontics, many cephalogrametric measurements have
been proposed as reliable indexes of facial attractive-ness. From the early work of Downs1 to the recent
work of Arnett et al,13 almost every proposed cephalo-metric analysis contains some measures of facial attrac-
tiveness. Hence, it seems reasonable to investigate the
concordance between the objective angular and lin-
ear measurements of x-ray cephalometry and the
subjective but highly reliable ranking of facial
photographs for attractiveness.
The underlying approach of this study was to use
such highly reliable but noninterval and hence only rel-
ative average rankings by clinicians of representative
Table III. Correlations between Photo Attractiveness Rank and 21 transformed cephalometric measures
US patients (n 5 45) Chinese patients (n 5 48)
Transformed value Transformed value
Order Cephalometric measure r P Difference* Order Cephalometric measure r P Difference*
1 Profile angle () 0.61 \0.0001 0.02 1 %Lower face height 0.44 0.002 0.35
2 H-angle () 0.53 0.0002 0.05 2 Chin prominence (mm) 0.43 0.002 0.01
3 Chin prominence (mm) 0.49 0.001 0.01 3 Angle of convexity () 0.38 0.008 0.07
4 Angle of convexity () 0.46 0.002 0.05 4 Z-angle () 0.37 0.009 0.12
5 Z-angle () 0.45 0.003 0.03 5 ANB () 0.36 0.01 0.01
6 Lower lip prominence
(mm)
0.40 0.008 0.02 6 Facial plane angle () 0.35 0.01 0.01
7 B-line to lower lip (mm) 0.37 0.01 0.18 7 H-angle () 0.33 0.02 0.07
8 SN-pogonion angle () 0.35 0.02 0.11 8 Mandibular plane angle
()
0.33 0.02 0.06
9 Upper lip prominence () 0.35 0.02 0.03 9 Nose prominence (mm) 0.31 0.03 0.12
10 Lower incisor
to A-pogonion line
(mm)
0.35 0.02 0.06 10 Profile angle () 0.28 0.06 0.04
11 B-line to upper lip (mm) 0.32 0.04 0.20 11 B-line to upper lip (mm) 0.21 0.16 0.19
12 Mandibular plane angle
()
0.28 0.06 0.10 12 Lower incisor
to mandibular
plane angle ()
0.19 0.19 0.05
13 E-line to lower lip (mm) 0.26 0.10 0.03 13 SN-pogonion angle () 0.19 0.21 0.09
14 Nasolabial angle () 0.25 0.13 0.00 14 Lower lip prominence
(mm)
0.18 0.22 0.01
15 ANB () 0.25 0.10 0.19 15 Lower incisor
to A-pogonion line
angle ()
0.10 0.48 0.23
16 Facial plane angle () 0.20 0.19 0.03 16 E-line to lower lip (mm) 0.10 0.49 0.24
17 Wits appraisal (mm) 0.15 0.31 0.06 17 Upper lip prominence
(mm)
0.08 0.58 0.02
18 %Lower face height 0.12 0.45 0.11 18 Nasolabial angle () 0.08 0.60 0.02
19 Nose prominence (mm) 0.12 0.45 0.13 19 Wits appraisal (mm) 0.04 0.79 0.02
20 Lower incisor to
mandibular
plane angle ()
0.05 0.74 0.02 20 B-line to lower lip (mm) 0.04 0.81 0.27
21 Lower incisor
to A-pogonion line
angle ()
0.04 0.79 0.01 21 Lower incisor
to A-pogonion line
(mm)
0.01 0.92 0.01
*Difference 5 change in r compared toTable II; %Lower face height in the Chinese sample had the largest increase in absolute r value and is the
only measure that increased sufficently after transformation to satisfy our criteria for high statistical significance. Other large increases in absolute r
values were found for B-line to lower lip and B-line to upper lip in the US sample, but these still did not meet our criteria for statistical significance.
American Journal of Orthodontics and Dentofacial Orthopedics Oh et al 762.e11Volume136,Number6
8/14/2019 1-s2.0-S0889540609009329-main
12/14
sets of facial photographs as a standard against which to
test the association of 21 widely used quantitative ceph-
alometric measures. The cephalometric variables cho-
sen for this test included classical hard-tissue
measures from Downs,1,2 Steiner,3 Riedel,20 and so
on, as well as soft-tissuemeasures from the morerecentanalyses of Holdaway,21 Arnett and Bergman,12 Arnett
et al,13 and Burstone.24 Separate assessments were
made for samples of treated Chinese and US orthodontic
patients.
Approximately one third of the 21 cephalometric
measures tested for patients of each ethnicity correlated
with the clinicians rankings for facial attractiveness for
at least 1 ethnicity with high statistical significance
(P\0.01). Chin prominence, angle of convexity,
H-angle, and Z-angle correlated significantly with
Photo Attractiveness Rank among patients of both US
and Chinese ethnicities. Some other measures correlatedsignificantly with Photo Attractiveness Rank among pa-
tients of 1 ethnicity but not of the other. Among the US
patients, these measures included profile angle, ANB an-
gle, and lower lip prominence. Among the Chinese pa-
tients, B-line to upper lip and mandibular plane angle
were included.
The statistically significant relationships described
in the previous paragraph were all linear; ie, facial at-
tractiveness tended to increase (or decrease) in a fairly
straight line throughout the full range of available
values for a particular measure. Additional data-trans-
forming tests were performed for all 21 cephalometric
measures for both ethnicities to detect any measures
for which clinicians preferred some ideal value in
the middle of the variable range, but values on either
side of the ideal tended to be found less attractive.We believed at the outset of this study that many ceph-
alometric measures would exhibit this pattern. But we
found this pattern to be of consequential importance
for only 1 measure, %lower face height among the Chi-
nese patients. For this variable, there was substantial ev-
idence that the judges considered values approximating
the ideal of 54% (Table I) to be most attractive, withvalues either above or below 54% tending to be consid-
ered less attractive. This implies that, among the Chi-
nese patients, the vertical orientation of the chin
contributed consequentially to the evaluation offacial
attractiveness (see the study of Johnston et al60 for
comparison).
The limited ability of our data-transforming proce-
dure to increase the number of statistically significant
correlations between the cephalometric measures and
the Photo Attractiveness Rank seems to imply that clini-
cians perception of attractiveness might not be as sym-
metrically distributed around the ideal values of most
cephalometric measures as has been thought. Almost
half of the 21 cephalometric measures commonly con-
sidered markers for facial attractiveness failed to corre-
late with Photo Attractiveness Rank better than chance
in either their raw or transformed form. Among the
poorer measures were E-line to lower lip, lower incisorto mandibular plane angle, and Wits appraisal. And even
among cephalometric measures that did correlate highly
with photographic rank, no single raw or transformed
measure had a higher correlation than 0.61, and no com-
bination of measures accounted for more than 41% of
the variability in the ranking of facial photographs forattractiveness by orthodontists.
In considering the relatively small portion of the to-
tal variance in Photo Attractiveness Rank accounted for
by the cephalograms, it should be recalled that the pho-
tographs used in the ranking process included frontal
and smiling views as well as a profile view. Recent stud-ies of facial photographs of the type examined here have
reported that frontal and frontal smiling images account
for a larger part of a judges conclusions about facial at-
tractiveness than do profile images.14,61,62 In general, it
seems fair to infer that most information that viewers
use in the evaluation of facial attractiveness is not avail-
able in lateral cephalograms.
As in any study, the limitations of the sample must
be considered when generalizing the study results.
The images in this sample were all taken at the end of
the orthodontic treatment of relatively attractive normal
Fig 5. A, Raw cephalometric values from Table II; B,
transformed values from Table III. The high statisticalsignificance of plot B shows that patients who had raw
%lower face height values close to the ethnic mean of
54% tended to have more favorable Photo Attractive-
ness Ranks than patients who had %lower face height
values either higher or lower than the ideal value.
762.e12 Oh et al American Journal of Orthodontics and Dentofacial OrthopedicsDecember 2009
8/14/2019 1-s2.0-S0889540609009329-main
13/14
subjects. The sample had been stratified to make it more
representative of the adolescent population seeking or-
thodontic treatment than of the general population. In
each sample at original presentation, half of the patients
were Class I, and half were Class II; half were treatedwith extractions, and half without extractions; three
quarters were girls, and a quarter were boys. These con-
straints might have tended to make the range of values
of the cephalometric measures smaller than would
have been the case for pretreatment cephalograms.
CONCLUSIONS
In general, correlations between cephalometric
measures and rankings of facial attractiveness were
less strong than had been expected. Soft-tissue profile
angle appeared to be the most important single variable
in accounting for differences in facial attractivenessrank on photographs of US patients, but it was not
nearly as important in ranking the photographs of Chi-
nese patients. In the Chinese patients, a combination
of Z-angle and transformed %lower face height contrib-
uted statistically significantly to the explanation of the
observed variability in Photo Attractiveness Rank.
However, no combination of cephalometric measures
for either ethnicity accounted for more than 42% of
the variance in Photo Attractiveness Rank, thus leaving
more than half of thevariance unexplained. In general, it
seems fair to infer that most of the information that
viewers use in the evaluation of facial attractiveness isnot available in lateral cephalograms.
For both Chinese and US patients, flatter facial pro-
files appeared to be preferred. In both ethnicities,
greater chin prominence in the sagittal direction was
preferred, and lip position was considered important.
With the exception of %lower face height in the Chinese
sample, close approximation to the ethnic ideal did
not appear to be highly correlated with facial attractive-
ness rank.
REFERENCES
1. Downs WB. Variations in facial relationship. Their significance in
treatment and prognosis. Am J Orthod 1948;34:812-40.
2. Downs WB. The role of cephalometrics in orthodontic case anal-
ysis and diagnosis. Am J Orthod 1952;38:162-82.
3. Steiner CC. Cephalometrics for you and me. Am J Orthod 1953;
39:729-54.
4. Tweed CH. Evolutionary trends in orthodontics, past, present, and
future. Am J Orthod 1953;39:81-108.
5. Tweed CH. The Frankfort-mandibular incisor angle (FMIA) in
orthodontic diagnosis, treatment planning and prognosis. Angle
Orthod 1954;24:121-69.
6. Jarabak JR, Fizzell JA. Technique and treatment with light-wire
edgewise appliance. 2nd ed. St Louis: C.V. Mosby; 1972.
7. Sassouni V. A classification of skeletal facial types. Am J Orthod
1969;55:109-23.
8. Bjork A. The face in profile. Svennsk Tandlakare-Tidskrift 40
No. 5B, suppl lund, 1947.
9. Ricketts RM. Perspectives in the clinical application of cephalo-
metrics. Angle Orthod 1981;51:115-50.10. Ricketts RM. Planning treatment on the basis of facial pattern and
an estimate of its growth. Angle Orthod 1957;27:14-37.
11. McNamara JA. A method of cephalometric evaluation. Am J
Orthod 1984;86:449-69.
12. Arnett GW, Bergman RT.Facial keysto orthodontic diagnosisand
treatment planning.Part I. Am J Orthod Dentofacial Orthop 1993;
103:299-31.
13. Arnett GW, Jelic JS, Kim J, Cummings DR, Beress A,
Worley CM, et al. Soft tissue cephalometric analysis: diagnosis
and treatment planning of dentofacial deformity. Am J Orthod
Dentofacial Orthop 1999;116:239-53.
14. Matoula S, Pancherz H. Skeletofacial morphology of attractive
and nonattractive faces. Angle Orthod 2006;76:204-10.
15. Tulloch C, Philips C, Dann C. Cephalometric measures as indica-
tors of facial attractiveness. Int J Adult Orthod Orthognath Surg1993;8:171-9.
16. Kiekens RMA, Maltha JC, Vant HofMA, Kuijpers-Jagtman AM.
Objective measures as indicators for facial esthetics in adoles-
cents. Angle Orthod 2006;76:551-6.
17. Xu TM, Korn EL, Liu Y, Oh HS, Lee KH, Boyd RL, et al. Facial
attractiveness: ranking of end-of-treatment facial photographs by
pairs of Chinese and US orthodontists. Am J Orthod Dentofacial
Orthop 2008;134:74-84.
18. Liu Y, Korn EL, Oh HS, Pearson H, Xu TM, Baumrind S.
Comparison of Chinese and US orthodontists averaged evalua-
tions of facial attractiveness from end-of-treatment facial
photographs. Am J Orthod Dentofacial Orthop 2009;135:
621-34.
19. Salzman JA. Roentgenographic cephalometrics. Philadelphia:
J.B. Lippincott; 1961.
20. Riedel RR. The relation of maxillary structures to cranium in
malocclusion and in normal occlusion. Angle Orthod 1952;22:
142-5.
21. Holdway RA. A soft tissue cephalometric analysis and its use in
orthodontic treatment planning. PartI. Am J Orthod 1983;84:1-28.
22. Holdaway RA. A soft tissue cephalometric analysis and its use in
orthodontic treatment planning. Part II. Am J Orthod
1984;85:279-93.
23. Ricketts RM. Cephalometric analysis and synthesis. Angle
Orthod 1961;31:141-56.
24. Burstone CJ. Integumental profile. Am J Orthod 1958;44:1-25.
25. Burstone CJ. Lip posture and its siginificance in treatment plan-
ning. Am J Orthod 1967;53:262-84.
26. Bell WH,Proffit WR,White RP. Surgicalcorrection of dentofacialdeformities. Vol I. Philadelphia: W.B. Saunders; 1980. p. 137-50.
27. Merrifield LL. The profile line as an aid in critically evaluating
facial esthetics. Am J Orthod 1966;52:804-22.
28. Jacobson A. The Wits appraisal of jaw disharmony. Am J
Orthod 1975;67:125-38.
29. Farkas LG. Anthropometry of the head and face in medicine. New
York: Elsevier; 1981.
30. Yu XH. A radiographic cephalometric study of soft-tissue profile
on 180 normal occlusions among Chinese population [thesis].
Beijing: Peking University; 1985.
31. Zhang XZ. Longitudinal study of craniofacial growth of 13-17
years adolescents with normal occlusion [thesis]. Beijing: Peking
University; 1995.
American Journal of Orthodontics and Dentofacial Orthopedics Oh et al 762.e13Volume136,Number6
8/14/2019 1-s2.0-S0889540609009329-main
14/14
32. Yuan DH, Zuo YP. The soft tissue profile analysis of normal oc-
clusion in the early permanent dentitiona cephalometric study.
Chinese Journal of Aesthetic Medicine 2007;1:95-8.
33. Lucker GW. Esthetics and a quantitative analysis of facial appear-
ance. In: McNamara JA Jr, editor. Esthetics and the treatment of
facial form. Monograph 28. Craniofacial Growth Series. AnnArbor, MI: Center for Growth and Development, University
of Michigan; 1993.
34. Edler RJ. Background considerations to facial aesthetics. J Orthod
2001;28:159-68.
35. Wahl N. Orthodontics in 3 millennia. Chapter 7: facial analysis
before the advent of the cephalometer. Am J Orthod Dentofacial
Orthop 2006;129:293-8.
36. Peck S, Peck L. Facial realities and oral esthetics. In:
McNamara JA Jr, editor. Esthetics and the treatment of facial
form. Monograph 28. Craniofacial Growth Series. Ann Arbor:
Center for Human Growth and Development; University of Mich-
igan; 1993.
37. Giddon DB, Sconzo R, Kinchen JA, Evans CA. Quantitative com-
parison of computerized discrete and animated profile prefer-
ences. Angle Orthod 1996;66:441-8.38. Orsini MG, Huang GJ, Kiyak AH, Ramsay DS, Bollen AM,
Anderson NK, et al. Methods to evaluate profile preferences for
the anteroposterior position of the mandible. Am J Orthod Dento-
facial Orthop 2006;130:283-91.
39. Ricketts RM. The biological significance of the divine proportion
and Fibonacci series. Am J Orthod 1982;81:351-70.
40. Olds C.Facial beautyin Westernart.In: McNamara JA, editor. Es-
thetics and the treatment of facial form. Monograph 28. Craniofa-
cial Growth Series. Ann Arbor: Center for Human Growth and
Development; University of Michigan; 1993.
41. Naini FB, Moss JP, Gill DS. The enigma of facial beauty: es-
thetics, proportions, deformity, and controversy. Am J Orthod
2006;130:277-82.
42. Euklid. The Elements, Books I-XIII. Darmstadt, Germany: Wis-
senschaftliche Buchgesellschaft; 1991.
43. Powell N, Humphries B. Proportions of the aesthetic face. New
York: C.M. Thieme-Stratton; 1984.
44. Clark K. Leonardo Da Vinci. Revised ed. New York: Penguin
Books; 1988.
45. Panofsky E. Durer as a theorist of art. In: The life and art of
Albrecht Durer. 4th ed. Princeton, NJ: Princeton University Press:
1955.
46. Gwilt J. The architecture of Marcus Vitruvius Pollio. London:
Priestly and Weale; 1826.
47. Alley TR, Cunningham MR. Averaged faces are attractive, but
very attractive faces are not average. Psychol Sci 1991;2:123-35.
48. Langlois JH, Roggman LA. Attractive faces are only average.
Psychol Sci 1990;1:115-21.49. Langlois JH, Roggman LA, Musselman L. What is average and
what is not average about attractive faces? Psychol Sci 1994;5:
214-20.
50. Newcombe RG. Averageness as the ideal. J Orthod 2002;29:76-8.
51. Henss R. Perceiving age and attractiveness in facial photographs.
J Appl Psychol 1991;21:933-46.
52. Mathes EW, Brennan SM, Haugen PM, Rice HB. Rating of phys-
ical attractiveness as a function of age. J Soc Psychol 1985;125:
157-68.
53. Grammer K, Thornhill R. Human (Homo sapiens) facial attrac-
tiveness and sexual selection: the role of symmetry and average-
ness. J Comp Psychol 1994;108:233-42.
54. Rhode G, Roberts J, Simmons L. Reflections on symmetry and at-
tractiveness. Psychology, Evolution, and Gender 1999;1:279-95.
55. Sarver DM. The importance of incisor positioning in the estheticsmile: the smile arc. Am J Orthod Dentofacial Orthop 2001;120:
98-111.
56. Sarver DM, Ackerman MB. Dynamic smile visualization and
quantification: part 1.Evolutionof theconcept anddynamicrecords
for smile capture. Am J Orthod Dentofacial Orthop 2003;124:
4-12.
57. Hulsey CM. An esthetic evaluation of lip-teeth relationships pres-
ent in the smile. Am J Orthod 1970;57:132-44.
58. Isxksal E, Hazar S, Akyalcn S. Smile esthetics: perception and
comparison of treated and untreated smiles. Am J Orthod Dento-
facial Orthop 2006;129:8-16.
59. Vanderburg N. The horizontal smile line in adolescent malocclu-
sions. IADR abstract 2583; 2004.
60. Johnston DJ, Hunt O, Burden CDE, Stevenson M, Hepper P. The
influenceof lower facevertical proportion on facial attractiveness.
Eur J Orthod 2005;27:349-54.
61. Shafiee R, Korn EL, Pearson H, Boyd RL, Baumrind S. Evalua-
tion of facial attractiveness from end-of-treatment facial photo-
graphs. Am J Orthod Dentofacial Orthop 2008;133:500-8.
62. Lundstrom A, Popovich F, Woodside DG. Panel assessments of
the facial frontal view as related to mandibular growth direction.
Eur J Orthod 1989;11:290-7.
762.e14 Oh et al American Journal of Orthodontics and Dentofacial OrthopedicsDecember 2009