ORIGINAL ARTICLE

Age-dependant cephalometric standards as

determined by multilevel modeling
Arjen F. van Diepenbeek,a Peter H. Buschang,b and Birte Prahl-Andersenc
Amsterdam, the Netherlands, and Dallas, Tex

Introduction: The purpose of this study was to evaluate the feasibility of constructing age-dependant
cephalometric standards for white subjects by using 3 data sets. Methods: The data sets were the samples
from the Fels Longitudinal Study (United States), the Michigan Growth Study (United States), and the
Nijmegen Growth Study (The Netherlands). The 3 mixed-longitudinal samples provided data for 218 girls and
231 boys between 9 and 14 years of age and were compared based on 4 cephalometric angles: SNA, SNB,
ANB, and SN/GoMe. Curve-fitting and statistical comparisons were performed with multilevel modeling
procedures. Results: All 4 angles showed linear changes over time. SNA and SNB increased, whereas ANB
and SN/GoMe decreased. Based on paired-sample comparisons, the samples displayed statistically
significant (P ⬍0.05) differences for 50% of the growth velocities and 8% to 17% of the intercepts (size of
the angle at 11 years). The SNA and SNB angles showed small and inconsistent differences across the
samples. The ANB angle for the Fels boys decreased less than in the other 2 samples. The Nijmegen and Fels
girls had the greatest and the least decreases, respectively, in the SN/GoMe angle. Most sample differences
decreased over time. Conclusions: Based on the growth differences identified, we concluded that sagittal
and vertical jaw relationships have different patterns of growth in different samples of white subjects;
indiscriminate pooling of data, to create age-dependant cephalometric standards for white subjects is not
recommended. (Am J Orthod Dentofacial Orthop 2009;135:79-87)

T
he ability to accurately predict and evaluate the
effects of orthodontic and surgical intervention
on facial growth has been attributed to our
knowledge and understanding of changes that take
place in treated and untreated subjects.1-3 Craniofacial
growth reference data (standards) play an essential role
in orthodontic diagnosis and treatment planning with
respect to timing, duration, and the amount and type of
orthopedic forces used. Reference data are also neces-
sary to determine the severity of the deficits of patients
with craniofacial growth disorders or congenital mal-
formations.
The applicability of cephalometric reference data is
largely determined by its external validity and its ability
to estimate extreme percentiles. We presently do not
have reference data that apply to all white subjects.
Reference data have been developed for separate
a
Private practice, Department of Orthodontics, Academic Center for Dentistry,
Amsterdam, The Netherlands.
b
Professor, Department of Orthodontics, Baylor College of Dentistry, Texas
A&M University Health Science Center, Dallas, Tex.
c
Chairperson, Department of Orthodontics, Academic Center for Dentistry,
Amsterdam, the Netherlands.
Reprint requests to: Arjen F. van Diepenbeek, Academic Center for Dentistry,
Louwesweg 1, 1066EA Amsterdam, the Netherlands; e-mail, diepenbeek@
hetnet.nl.
Submitted, January 2006; revised and accepted, November 2006.
0889-5406/$36.00
Copyright © 2009 by the American Association of Orthodontists.
doi:10.1016/j.ajodo.2006.11.025
mixed-longitudinal data sets, but differences between
samples are largely untested.4-9 To obtain accurate
estimates of extreme percentiles, large sample sizes are
required. Because no data sets are sufficiently large to
estimate extreme percentiles and it is unlikely (due to
costs and ethical considerations) that larger longitudinal
studies will be conducted in the future, we must use the
existing longitudinal data fully and efficiently.
To that end, the purpose of this study was to
evaluate the feasibility of constructing age-dependant
cephalometric standards for white subjects. By combin-
ing 3 large data sets—the Fels Longitudinal Study
(United States),5 the Michigan Growth Study (United
States),6 and the Nijmegen Growth Study (The Neth-
erlands)7— our long-term goal was to derive reference
data flexible enough to be applied to white subjects
generally and sufficiently precise to identify abnormal
growth patterns. Mathematical models will be devel-
oped so that the reference data can be easily incorpo-
rated into existing cephalometric programs.
MATERIAL AND METHODS
The data were derived from 3 large mixed-longitu-
dinal cephalometric data sets, including samples from
the Fels Longitudinal Study, the Michigan Growth
Study, and the Nijmegen Growth Study. The age range,
9 to 14 years, was chosen for both girls and boys
because data were available from all 3 sets in this
79
80 van Diepenbeek, Buschang, and Prahl-Andersen American Journal of Orthodontics and Dentofacial Orthopedics
January 2009

Table I. Total subjects per data set Table II. Total measurements per site and age
Girls Boys Total per
9 10 11 12 13 14 site
Fels 46 47
Michigan 82 98 Girls
Nijmegen 90 86 Fels 39 31 42 35 41 32 220
Total 218 231 Michigan 56 58 54 42 40 44 294
Nijmegen 155 162 135 89 79 21 641
Total per age 250 251 231 166 160 97 1155
Boys
range, and children are usually treated between those Fels 29 22 38 36 44 39 208
Michigan 63 76 56 63 66 51 275
ages. Nijmegen 149 152 114 67 47 18 547
The sample from the Fels study included children Total per age 241 250 208 166 157 108 1130
born between 1959 and 1969 living in southwestern
Ohio. The radiographs were taken annually between 4
and 18 years of age. A total of 562 cephalograms,
pertaining to a mixed-longitudinal sample of both
sexes, were digitized after tracing. Children who had
orthodontic treatment before or during the collection
process were excluded from this study. Nonwhite
subjects (3%) in the Fels sample were excluded. A total
of 46 girls and 47 boys between 9 and 14 years of age
were included.
The data for the Michigan Growth Study came
from the Elementary and Secondary School Growth
Study at the University of Michigan and were col-
lected between 1955 and 1974. One hundred eighty
subjects with continuous attendance at the university
over the period from 6 to 16 years of age were
selected from the total sample. The radiographs were
taken on birthdays during the school year or on the
subject’s half-birthday if the birthday was outside the
school year. The data were exclusively taken from
Fig 1. Cephalometric landmarks.
lateral cephalograms of orthodontically untreated
children. All radiographs were traced and digitized.
A total of 82 girls and 98 boys were included in this
study from 9 to 14 years of age. each data set. Table II shows the total number of
The Nijmegen Growth Study began in 1970 in the measurements per site and age.
Netherlands to study the interrelationships among The 3 samples included untreated white subjects of
various developmental processes. The cephalometric middle to northern European ancestry. Only children
data came from existing mixed-longitudinal records with normal occlusion and untreated malocclusion
taken between 1971 and 1976, by using 6 overlap- (dental and skeletal) were included in the study. Each
ping cohorts. At the start of the study, the children sample represents its respective ethnic group; geo-
were 4, 4.5, 7, 7.5, 9.25, or 9.5 years of age. graphic and cultural variations exist between the 3
Radiographs were taken annually from 4 to 9 years samples.
and semiannually from 9 to 14 years of age. About The comparisons of the 3 groups were based on 4
2500 lateral cephalometric radiographs, pertaining to craniofacial measurements representing sagittal and
467 subjects (220 girls, 247 boys) were digitized vertical jaw relationships: SNA, SNB, ANB, SN/GoMe
directly from the cephalograms and stored for anal- angles. The 4 measurements were computed from 6
ysis. Data from 90 girls and 86 boys 9 to 14 years of digitized landmarks (Fig 1).
age were included in this study. The number of Cephalometric landmarks used include sella tur-
measurements exceeds the number of persons be- cica (S), the optic geometric midpoint of the sella
cause more than 1 cephalogram was available at each turcica; nasion (N), the midpoint of the naso-frontal
age. Table I shows the number of girls and boys in suture lying on the outer contour of the bony fore-
American Journal of Orthodontics and Dentofacial Orthopedics van Diepenbeek, Buschang, and Prahl-Andersen 81
Volume 135, Number 1

Table III. Total measurements, missing data, and left out data
SNA SNB ANB SN/GoMe

F M N F M N F M N F M N

Girls
Subjects 220 294 641 220 294 641 220 294 641 220 294 641
Missing 0 0 6 0 0 8 0 0 11 0 0 11
Left out 3 5 3 1 2 3 3 5 3 0 2 2
Total 217 289 632 219 292 630 217 289 627 220 292 628
Boys
Subjects 208 375 547 208 375 547 208 375 547 208 375 547
Missing 0 0 6 0 0 3 0 0 8 0 2 5
Left out 3 3 3 3 3 1 3 3 3 2 3 3
Total 205 372 538 205 372 543 205 372 536 206 370 539

F, Fels; M, Michigan; N, Nijmegen.

head; A-point (A), the deepest point on the concave
curvature between the anterior nasal spine (ANS),
and prosthion (Pro), defined as the crest of the
alveolar process of the most prominent maxillary
central incisor and the ANS as the foremost maxil-
lary point in the region of the nasal floor; B-point
(B), the deepest point on the concave curvature
between infradentale (Inf) and pogonion (Pog), with
infradentale defined as the crest of the alveolar
process of the most prominent mandibular incisor
and pogonion as the foremost point of the outer
contour of the chin; menton (Me), the lowest point on
the outer contour of the chin; gonion (Go), the
intersection of the outer contour of the mandible with
the bisector of the angle formed by a tangent to the
posterior ramus from articulare (Ar).
The magnification factors were 9.09% for the Fels
Longitudinal Study, 12.9% for the Michigan Growth
Study, and 4.17% for the Nijmegen Growth Study.
Because only angular measurements were involved,
magnification was not adjusted.10
The Michigan and Nijmegen radiographs had been
previously digitized. The Fels data base was traced and
digitized by Riesmeijer et al.11 They reported intraex-
aminer cephalometric tracing reliabilities (intraclass
correlations) from 0.988 to 0.999; the reliability of
digitization ranged from 0.989 to 1.000.
Statistical analysis
For each measurement, a check was made for
outliers. Outliers were defined as a fluctuation of
sequential measured values of more than 3°. For
example, if SNA at 9 years is 80°, at 10, 84°, and at 11,
81°, the measurement at 10 years (84°) was considered
an outlier and therefore left out.
Table III shows the total amount of measurements,
missing data, and left out data.
Once the data had been checked and cleaned, the
age changes of the 4 measurements were modeled by
using MLwiN (version 1.1, Centre for Multilevel Mod-
elling, Institute of Education, University of London,
London, United Kingdom). Multilevel modeling proce-
dures, as described by Goldstein,12 can easily handle
missing data and do not require adjustment of measure-
ments to exact ages. The multilevel models partitioned
random variation within and between subjects. The
procedure first evaluated the order of the polynomial
necessary to fit all the subjects’ data. So the model
represents a sort of mathematical function, which de-
scribes the curve that best fits the measures. Multilevel
modeling creates a mathematical function that best fits
all measurements or the measurements of a group of
persons at different ages.
The fixed part of the multilevel model describes the
mean growth curves. Polynomials were used to model
each measure as a function of age. Multilevel is a
powerful tool to create growth curves and has been
extensively used in orthodontic research.13-15
For numeric stability, we subtracted 11 from each
age, thereby transforming the age range from 9 to 14 to
–2 to 3. The polynomial represents the mathematical
function of the curve that best fits the measurements.
For example, if the growth curve is linear with a
positive slope, the formula for the estimate at 10 years
of age would be Y(age 10) ⫽ Con ⫹ (slope * –1). The
constant (con) would be the mean size at the intercept
(ie, at 11 years of age), and the slope is the increase or
decrease in Y per year in age. In other words, it
provides a measure of growth velocity at 10 years of
age. Once the model is established, it is possible to
calculate the measure’s values at every age in the age
span modeled.
The 3 samples were first analyzed separately to
determine the appropriate models for the SNA, SNB,
82 van Diepenbeek, Buschang, and Prahl-Andersen
Fig 2. SNA for both sexes.
ANB, and SN/GoMe angles. Separate models were
estimated for the sexes. Paired sample comparisons
were then made to statistically determine differences
(intercept and slope) among the 3 samples.
RESULTS
All 4 angles changed as a linear function of age
between 9 and 14 years of age (Figs 2-5). With the
exception of the Michigan female SNA angle and the
Fels male ANB angle, all changes were statistically
significant (P ⬍0.05). SNA and SNB increased by 0.1°
to 0.3° and 0.2° to 0.4° per year, respectively. ANB
angle decreased by 0.1° to .02° per year, and SN/GoMe
American Journal of Orthodontics and Dentofacial Orthopedics
January 2009
angle decreased by 0.2° to 0.6° per year. Of the 4
measurements, SN/GoMe showed the greatest yearly
change (Fig 5).
Table IV gives the sex-specific multilevel models
by sex (intercept [SE intercept] ⫹ slope [SE slope]).
Linear functions were used to estimate values
between 9 and 14 years of age for each measurement
(Figs 2-5). For example, SNA for the girls was 80.83°
at 11 years of age, 80.68° (80.83 ⫹ 0.15 [X⫺1]) at 10
years of age and 81.13 (80.83 ⫹ 0.15 [X2]) at 13 years
of age (Table IV and Fig 2).
The paired statistical comparisons showed that
12.5% of the intercepts and 50% of the slopes were
American Journal of Orthodontics and Dentofacial Orthopedics
Volume 135, Number 1
Fig 3. SNB for both sexes.
significantly (P ⬍0.05) different across the samples
(Figs 2-5).
For the girls, there were no significant sample differ-
ences in size at age 11 for SNA, SNB, and ANB. The
SN/GoMe angle was significantly larger for the Nijmegen
than the Michigan girls. Girls had significant sample differ-
ences in growth velocity for SNA, SNB, and SN/GoMe.
Increases in the SNA and SNB angles tended to be the
greatest for the Nijmegen girls and the least for the Michigan
girls. Decreases in the SN/GoMe angle were greatest for
the Nijmegen girls and least for the Michigan girls.
van Diepenbeek, Buschang, and Prahl-Andersen 83
For the boys, there were no significant size
differences at age 11 for SNA, SNB, and SN/GoMe.
The ANB angle of the Fels boys was significantly
smaller than that of the other 2 male samples. Boys
had significant differences in growth velocity for all
measurements except SN/GoMe. As was evident in
the girls, SNA and SNB angles increased most for
the Nijmegen boys and least for the Michigan boys.
The Fels boys had significantly less reduction in
ANB angle than did the boys from Michigan and
Nijmegen.
84 van Diepenbeek, Buschang, and Prahl-Andersen
Fig 4. ANB for both sexes.
DISCUSSION
Children are usually treated between 9 and 14 years
of age; therefore, this age span was chosen for the
construction of reference data for craniofacial growth.
The SNA and SNB angles increased by 0.2° to 1.3° and
1° to 1.8°, respectively, between 9 and 14 years of age,
indicating greater relative sagittal development of the
mandible than the maxilla. Similar increases were
reported for children from Philadelphia, Pa (Walker
and Kowalsky16) and Denver, Colo (Banafsheh and
Nanda17). Bishara,18 following a sample from 5 to 25
years of age, found that 78% of the total increase of
American Journal of Orthodontics and Dentofacial Orthopedics
January 2009
SNA occurred between 10 and 15 years of age for boys.
For girls, 50% of the increase occurred between 10 and
15 years of age. Bishara8 showed an increase in SNA
for boys (from 80° at 8 years to 81° at 12), but not for
girls between 8 and 14 years of age. However, he
reported SNB increases of 2° for boys and 1° for girls.
El-Batouti et al9 reported somewhat larger increases of
SNA (1.8°-2.6°) and SNB (2.1°-2.4°) for boys and girls
9 to 15 years of age. Our results are also consistent with
previous reports showing greater increases in SNA and
SNB for boys than girls.8,9,17 Boys have relatively
greater anterior displacements of the midface and the
American Journal of Orthodontics and Dentofacial Orthopedics
Volume 135, Number 1
Fig 5. SN/GoMe for both sexes.
mandible than do girls; these might be related to
well-established differences in growth rates17,19 and
differences in maxillary and mandibular rotations.20
The age changes in the ANB angle were due to
differences in maxillary and mandibular growth, with
girls showing greater decreases than boys. Bishara et
al21 showed that the ANB angles of boys and girls
decreased 0.1° and 0.9°, respectively; this compares
closely with our results. In 2000, Bishara18 also found
a decrease of ANB in his population of Iowa school
children. Six percent of the total decrease between 5
and 25 years of age took place between ages 10 and 15
van Diepenbeek, Buschang, and Prahl-Andersen 85
years, 63% after 15 years, and 31% before 10 years for
the boys. The total decrease of ANB from 5 to 25 year
was 1.6°. For the girls, 71% the decrease occurred
between 10 and 15 years of age, with a total decrease of
1.4° from 5 to 25 years of age.
Nanda and Ghosh,22 in a longitudinal study of the
sagittal positions of the maxilla and mandible, showed
that Point B moved forward more than Point A between
6 and 24 years of age. Girls had greater forward
movement from ages 6 to 12; boys had greater forward
movement from 12 to 18 years. This suggests that the
sex differences we observed in ANB decrease can be
86 van Diepenbeek, Buschang, and Prahl-Andersen
Table IV. The sample and sex-specific multilevel
models*
Girls Boys
SNA
Fels 80.83 (0.48) ⫹ 0.15 (0.05) 79.38 (0.49) ⫹ 0.24 (0.06)
Michigan 80.27 (0.38) ⫹ 0.04 (0.04) 80.41 (0.32) ⫹ 0.13 (0.03)
Nijmegen 80.36 (0.36) ⫹ 0.21 (0.03) 80.30 (0.38) ⫹ 0.27 (0.03)
SNB
Fels 72.24 (0.49) ⫹ 0.35 (0.05) 76.80 (0.41) ⫹ 0.26 (0.05
Michigan 76.49 (0.35) ⫹ 0.20 (0.04) 76.28 (0.30) ⫹ 0.25 (0.03)
Nijmegen 76.31 (0.33) ⫹ 0.40 (0.03) 76.08 (0.36) ⫹ 0.37 (0.03)
ANB
Fels 3.61 (0.24) ⫺ 0.20 (0.03) 2.59 (0.26) ⫺ 0.02 (0.04)
Michigan 3.78 (0.26) ⫺ 0.16 (0.03) 4.13 (0.19) ⫺ 0.12 (0.03)
Nijmegen 4.05 (0.20) ⫺ 0.19 (0.03) 4.21 (0.22) ⫺ 0.10 (0.02)
SN/GoMe
Fels 36.03 (0.74) ⫺ 0.43 (0.06) 34.43 (0.70) ⫺ 0.28 (0.07)
Michigan 34.57 (0.55) ⫺ 0.19 (0.05) 34.67 (0.48) ⫺ 0.32 (0.04)
Nijmegen 36.54 (0.51) ⫺ 0.57 (0.04) 35.31 (0.48) ⫺ 0.42 (0.04)
*Intercept (SE intercept) ⫹ slope (SE slope).
partially explained by maturity differences, with girls
well into their adolescent spurt.
The mandibular plane angle decreased between 9
and 14 years of age, indicating that mandibular growth
direction changed toward a more sagittal direction. Sex
differences in Sn/GoMe become smaller with age. Van
der Beek et al23 previously reported a decrease of the
mandibular plane angle between 7 and 14 years of age
for Dutch girls. Buschang et al20 showed that the
growth direction during childhood and adolescence was
predominately horizontal; this might be associated with
greater forward mandibular rotation. In terms of remod-
eling, decreases in the mandibular plane angle and
forward mandibular rotation have been related to in-
creased vertical growth of the posterior mandible and
increased resorption in the gonial region.24,25 Impor-
tantly, forward mandibular rotation also affects sagittal
jaw relationships and helps to explain the decreases
observed for the ANB angle.
The differences between the 3 samples of white
subjects were inconsistent. The ANB angle at 11 years
of age was smaller for th Fels boys than for the
Michigan and Nijmegen boys because SNA was
slightly smaller and SNB was slightly larger. For the
girls, there was a significant difference of 1.98° in
SN/GoMe between the Nijmegen and the Michigan
samples. Importantly, these differences decreased over
time. The Nijmegen and Michigan girls showed slightly
larger and smaller increases in SNA and SNB angles,
respectively, than did the Fels sample. Most (75%) of
the yearly changes were less than .2°, and all but 1 were
less than .3° per year. Trenouth et al26 previously
American Journal of Orthodontics and Dentofacial Orthopedics
January 2009
showed that the SNA and SNB angles of the Nijmegen
and Michigan children were comparable at 9, 10, and
11 years of age; this supports our findings. However,
those authors also showed that both samples had larger
SNA and SNB angles than the Manchester children;
they attributed this to possible differences in the eth-
nicity of the white samples. The group differences are
probably not due to our lack of control over Angle
classifications, because Riesmeijer et al,11 who evalu-
ated the same sample, found no significant differences
for either Class I or Class II subjects.
The results indicate that the samples could be
pooled for some measurements, but probably not for
others. Because the differences were generally less than
1°, it might be acceptable to pool the samples for the
SNA and SNB angles. Normal variability for SNA and
SNB is 3° to 4°, indicating that one would expect to
find differences in any sample within 12° to 16° (SD, ⫾
1.96°).3,6,8,9 On that basis, the differences among the
Michigan, Fels, and Nijemegen samples represent less
than 8% of the variation normally seen in any sample.
In contrast to the SNA and SNB angles, the group
differences for the ANB and SN/GoMe angles were too
large in the younger ages to allow pooling without
adjustments. Although the differences decrease to less
than 1° at 14 years for SN/GoMe, the Fels ANB angle
remained 1.2° to 1.4° smaller than the Michigan and
Nijmegen ANB angles at the same age. These differ-
ences are systematic and, as such, are easily controlled
by making appropriate age, sex, and sample adjust-
ments.
CONCLUSIONS
1. Growth changes in sagittal and vertical jaw rela-
tionships showed significant differences between
samples of untreated white children, aged 9 to 14
years, from the Fels, Michigan, and Nijmegen
studies.
2. Indiscriminate pooling of cephalometric data from
these 3 samples to create age-dependant cephalo-
metric standards is not recommended.
3. Caution should be exercised when using historic
controls because children from other countries and
regions within countries might be expected to
mature at different rates and ages.
We thank A. M. Riesmeijer and K. W. L. Vig for
allowing us to use their data, especially after all their
work for their research published in the American
Journal of Orthodontics and Dentofacial Orthopedics
in 2004.
American Journal of Orthodontics and Dentofacial Orthopedics
Volume 135, Number 1
REFERENCES
1. Björk A. Facial growth in man, studied with aid of metallic
implants. Odontol Scand 1955;13:9-34.
2. Enlow DH. Handbook of facial growth. Philadelphia: Saunders;
1982.
3. Broadbent BH Sr, Broadbent BH Jr, Golden WY. Bolton
standards of dentofacial development growth. St Louis: C. V.
Mosby; 1975.
4. Hunter WS. An inventory of United States and Canadian growth
record sets: preliminary report. Am J Orthod Dentofacial Orthop
1993;103:545-55.
5. Roche AF. Growth, maturation and body composition. The Fels
longitudinal study, 1929-1991. Cambridge, Cambridge Univer-
sity Press; 1992.
6. Riolo ML, Moyers RE, McNamara JA, Hunter WS. An atlas of
craniofacial growth. Monograph 2. Ann Arbor: Center for
Human Growth and Development; University of Michigan;
1974.
7. Prahl-Andersen B, Kowalski CJ, Heyendael P. A mixed-longi-
tudinal interdisciplinary study of growth and development. New
York: Academic Press; 1979.
8. Bishara SE. Longitudinal cephalometric standards from 5 years
of age to adulthood. Am J Orthod 1981;79:35-44.
9. El-Batouti A, Øgaard B, Bishara SE. Longitudinal cephalometric
standards for Norwegians between the ages of 6 and 18. Eur
J Orthod 1994;16:501-9.
10. Dibbets JMH, Nolte K. Effect of magnification on lateral
cephalometric studies. Am J Orthod Dentofacial Orthop 2002;
122:196-201.
11. Riesmeijer AM, Prahl-Andersen B, Macarenhas AK, Joo BH,
Vig KWL. A comparison of craniofacial Class I and Class II growth
patterns. Am J Orthod Dentofacial Orthop 2004;125:463-71.
12. Goldstein H. Multilevel models in educational and social re-
search. London: Oxford University Press; 1987.
13. Gilthorpe MS, Cunningham SJ. The application of multilevel,
multivariate modelling to orthodontic research data. Community
Dent Health 2000;17:236-42.
van Diepenbeek, Buschang, and Prahl-Andersen 87
14. Hoeksma JB, van der Beek MC. Multilevel modelling of longi-
tudinal cephalometric data explained for orthodontics. Eur
J Orthod 1991;13:197-201.
15. Buschang PH. Mandibular growth prediction: mean growth
increments versus mathematical models. Eur J Orthod 1990;12:
290-6.
16. Walker GF, Kowalski CJ. Use of angular measurements in
cephalometric analyses. J Dent Res 1972;51:1015-21.
17. Banafsheh KO, Nanda RS. Comparison of maxillary and mandib-
ular growth. Am J Orthod Dentofacial Orthop 2004;125:148-59.
18. Bishara SE. Facial and dental changes in adolescents and their
clinical implications. Angle Orthod 2000;79:471-83.
19. Buschang PH, Tanguay R. A polynomial approach to craniofa-
cial growth: description and comparison of adolescent males
with normal occlusion and those with untreated Class ⌱⌱ maloc-
clusion. Am J Orthod Dentofacial Orthop 1986;90:437-42.
20. Buschang PH, Tanguay R, Demirjian A, LaPalme L, Goldstein
H. Modeling longitudinal mandibular growth: percentiles for
gnathion from 6 to 15 years of age in girls. Am J Orthod
Dentofacial Orthop 1989;95:60-6.
21. Bishara SE, Fahl JA, Peterson LC. Longitudinal changes in the
ANB angle and Wits appraisal: clinical implications. Am J
Orthod 1983;84:133-9.
22. Nanda RS, Ghosh J. Longitudinal growth changes in the sagittal
relationship of maxilla and mandible. Am J Orthod Dentofacial
Orthop 1995;107:79-90.
23. Van der Beek MC, Hoeksma JB, Prahl-Andersen B. Vertical
facial growth: a longitudinal study from 7 to 14 years of age. Eur
J Orthod 1991;13:202-8.
24. Buschang PH, Gandini LG Jr. Mandibular skeletal growth and
modelling between 10 and 15 years of age. Eur J Orthod
2002;24:69-79.
25. Baumrind S, Ben-Bassat Y, Korn EL, Bravo LA, Curry S.
Mandibular remodeling measured on cephalograms. Am J
Orthod Dentofacial Orthop 1992;102:134-42.
26. Trenouth MJ, Davies PHJ, Johnson SJ. A statistical comparison
of 3 sets of data from which to derive standards for craniofacial
measurement. Eur J Orthod 1985;7:193-200.