ARTICLES

metric supetimposition on the cranial b

rev&xv and a comparison of four metho
Joseph Ghafari,* Francoise E. Engel,** and Larry L. Laster***
PhiZadelphia,Pa., and San Antonio, Texas

Spatial change in the jaws of growing persons is often evaluated by superimposing cephalometric
tracings made at different points in time. Methods of superimposition vary according to structures
used as references within the skull. This study compares four different superimposition methods.
The sample consisted of 26 patients (13 boys, 13 girls) treated for Class II, Division 1 malocclusions
with extraction of the four first premolars. Tracings of pretreatment (average age for boys, 12.5
years; for girls, 12.2 years) and posttreatment (average age for boys, 15.4 years; for girls, 14.9
years) cephalograms were superimposed according to the following methods: (1) best fit of anterior
cranial base anatomy, (2) superimposition on SN line, registered at S, (3) superimposition on
registration point R with Bolton-nasion planes parallel, and (4) superimposition on basion-nasion
(Ricketts), registered at point CC (4) and point N (4a). Differences in amount of change among
the superimposition methods were assessed independently for each of the following landmarks: PNS,
ANS, A, B, Pog, Gon. On each patient and for each landmark, ten distances-the paired d~ffere~c~~
of five posttreatment positions obtained by methods 1, 2, 3, 4, and 4a-were evaluated. Two
methods were compared at a time. A t test examined the average difference for each comparison,
Because all differences between all paired methods were significant (P < 0.01) t tests were then
viewed under the hypothesis that a difference <I mm was insignificant clinically. Clinically-statistically
significant differences were found only for boys and for the total sample between methods 4a and
each of methods 1, 2, and 3. As method 4a is advocated to assess changes of point A (Rickets),
this method gives, for the same person, an interpretation of anterior maxillary change in position
different from the other methods. Conclusions about facial changes may be made only in reference
to the superimposition method. (AM J ORTHOD DENTOFAC ORTHOP 1987;91:403-13.)

Key words: Cephalometrics, superimposition, facial change, cranial reference

REVIEW OF THE LITERATURE

Spatial change of craniofacial structures is usually
evaluated by superimposition of cephalometric tracings
taken at different times. Methods of superimposition
differ according to reference structures used within the
skull. To evaluate facial changes, such structures must
be relatively stable over time and located outside the
face itself. Unlike the facial skeleton, the cranial base
grows rapidly in early postnatal life, reaching 90% of
its final size by 4 to 5 years of age.‘,’ For this reason,
the cranial base is considered a satisfactory reference
for cephalometric superimpositions.3.4
From the University of Pennsylvania School of Dental Medicine.
*Assistant Professor of Orthodontics.
**Formally Clinical Associate of Orthodontics (current address: University of
Texas Health Science Center, School of Dentistry).
***Associate Professor of Biostatistics, University of Pennsylvania, School of
Veterinary Medicme.
Four major methods of superimposition have been
reported.
1. Superimposition on the best jit of the anterior
cranial base anatomy. This method is based on de Cos-
ter’s observation of a stable basicranial line from the
inner contour of the frontal bone to the anterior aspect
of the sella turcicaS5 BjBrk3T4advocated superimposing
the anterior wall of the sella turcica, the anterior con-
tours of the middle cranial fossae, the contours of the
cribriform plate and frontoethmoidal crests, and pos-
sibly the cerebral surfaces of the orbital roofs and the
cortical layers of the frontal bone. Details of the tra-
beculations within the cranial structures provide further
guidance.5
This method is in agreement with Ford’s findings
that the cribriform plate stops growing in length an-
teroposteriorly after 2 years of age, and that the distance
from pituitary point to foramen caecum stops growing
in length after 6 to 7 years of age.6
The method is also consistent with Melsen’s obser-
 4 Ghafari, Engel, and Laster
Fig. 1. Superimposition at registration point R with Bolton-na-
sion planes of pretreatment (dotted line) and posttreatment
(solid line) tracings kept parallel.
Fig. 2. A, Superimposition on basion-nasion plane, registered
at point CC. B, Superimposition on basion-nasion plane, reg-
istered at nasion.
vations that, with few exceptions, the internal surface
of the frontal bone and the cribriform plate are stable
after the age of 6 to 7 years in both sagittal and vertical
planes7 Melsen reported that the anterior part of sella
turcica is by far the most stable over 5 years of age.
However, because of the remodeling in the sella turcica
region, the reference sella is not regarded as stable until
long after puberty.
2. Superimposition on sella-nasion (SN). SN is a
frequently used reference line2*8-‘6that has been reported
to be relatively stable.” Both points S and N are located
in the midsagittal plane and are displaced a minimal
degree by movement of the head. Steiner used SN with
registration point at sella to evaluate sagittal changes in
mandibular positions and at nasion to evaluate the po-
sition of the maxilla through changes in the angle
SNA.” Unlike Steiner, Bjijrk used sella as registration
Am. J. Qrthod. Dent&c. Orthop.
May 1984
Fig. 3. Coordinate system and cephatometric landmarks: nasion
(N), sella (S), anterior nasal spine (ANS), posterior na.sal spine
(PNS), A, B, pogonion (Pog), gonion (Gon), and articuiare (Ar).
point to assess changes in position of both jaws.14 He
advocated the use of SN as a particularly suitable ref-
erence during adolescence because of the constancy
(90% of cases) in the relation between SN and the
deepest median contour of the anterior cranial fossa.
Later, BjGrk reported that errors of biologic origins of
S and N may weaken the SN reference for estimation
of facial changes.3,4He stated that an upward or down-
ward displacement of nasion may occur with growth at
the frontonasal suture. Likewise, a posterior displace-
ment of sella may be induced by the remodeling of
dorsum sellae connected with the increased size of the
pituitary gland.
Strammd advocated the use of sella-ethmoidale
(SE) because of the variation in direction of growth of
nasion. However, SN and SE correlated closely from
3 years to adult age. l8
Ricketts, Schulhof, and Bagha suggested that the
Frankfort horizontal (FH) is a more reliable reference
than SN because the correlation between the measure-
ments of the maxilla and the mandible to FH (FH-NA
and FH-NPog) was minimal. Compared with the cor-
relation between the measurements to SN (SNA and
SNB). I9 On the other hand, in a comparative study of
five commonly used cephalometric reference lines, Wei
concluded that SN was less variable than FH.”
3. Superimposition at re~istr~tion~oi~t~ with Bol-
ton-nasionplarzesparallel. This method was introduced
Volume 91
Number 5
le 1.Distribution of patients, tracings, and correspondingnumber of superimpositions
Eloys 13 26 13
Girls g 26 13
TOTAL 26 52 26
by Broadbent.20~2’ A perpendicularto sella (S) is erected
from the Bohon-nasion(Bo-N) plane. The midpoint of
this perpendicular is called the registration point R.
Tracingsof successivecephalographsare superimposed
on R, keepingBo-N planesof theseradiographsparallel
to eachother (Fig. 1). Broadbentbasedthis methodon
observationsof dried skulls and a comparativestudy
of cranial base planes (Bolton-nasion, porion-nasion,
sella-nasion)in persons3 to 18 yearsof age. Although
statisticalcomparisonsamongthe different groupsstud-
ied were not done, the coefficient of variability was the
smallest for Bo-N plane.
4. Superimposition on basion-nasion plane. This
methodwas advocatedby Ricketts22who, like Steiner,12
usestwo different registrationpoints. After tracing Ba-
N plane, a point called pterygoid point (Pt) is selected
on the lower rim of the foramenrotundumas the highest
and most posterior point of the pterygopalatinefossa.
A line from Pt to cephalometricgnathion (Gn), which
is located at the intersectionof the facial plane and the
mandibular plane, constitutesthe central axis. At the
intersection of Ba-N and the central axis is located a
point CC, which is used as a referencecenter.
The angle formed by Ba-N and the central axis
indicates the position of the mandible relative to the
cranial base. By superimposingon Ba-N, registeredat
CC, the direction of growth of the mandibleis evaluated
by changesin the direction of the central axis (Fig. 2,
A). Registration at nasion depicts the changesin po-
sition of the maxilla through the movementof point A
(Fig. 2, B). The angle Ba-N-A would be the equivalent
of the angle SNA of Steiner.
Basedon studiesof laminographsections,Ricketts
suggestedthat the cranial base angle, while constant
on average,exhibits a changeof 5” in either direction
over a 3-yearperiod.23He concludedthat the use of Ba
instead of S, “helps take into account some of the
extremesin divergent growth conditions.” He consid-
ered Ba-N plane as a line of separationof the face from
the skull and hencea basic cranial axis for growth and
structural reference.
Nevertheless,one may doubt the reliability of this
13 13 13 13
r3 13 12 13
26 26 26 26
axis becausegrowth at nasionis subjectedto individual
variations. Moreover, the position of basion is inllu-
enced by remodeling processeson the clivus surface
and on the anteriorborderof the foramenmagnum, and
by changesin the position of the pars basilaris ossis
occipitalis associatedwith growth in the sp~en~-o~cip-
ital synchondrosis.7
O ther methods of superimpositionhave been de-
scribedusing referenceswithin the anteriorcranial base,
the posterior base, the upper face, or combinationsof
the previousreferences.17z4-29
The four methods of superimposition discussed
herein have been most commonly used without a def-
inite knowledgeof their respectivevalidity. The present
study comparesthesemethodsto evaluatetheir differ-
ences or similarities when interpreting facial changes
within the sameperson.
MATERIAL AND METHODS
The sample consistedof 26 patients (13 boys and
13 girls) treatedby the sameorthodontist(Brainerd F.
Swain) for ClassII, Division 1 malocclusionsincluding
extraction of the four first premolars. Patients were
selectedat random from a larger sample of similarly
treatedmalocclusions.
Treatmentin boys startedat an averageage of 12.5
years (11 years to 14 years) and was finished at an
average age of 15.4 years (14 years to 16.5 years);
treatmentin girls startedat an averageageof 12.2 years
(10 yearsto 15.5 years)and was finished at an average
age of 14.9 years (13 years to 18 years).
Radiographiccephalogramswere taken before and
after treatmentwith the samecephalostat.Theseradio-
graphswere tracedby one operator(FE.) and the trac-
ings superimposedaccordingto the following methods:
- Method 1: Best fit of the anteriorcranial baseanat-
omy-that is, the areabetweenthe anteriorclinoid
processand the frontoethmoidal suture
- Method 2: Superimpositionon SN line, registered
at S
- Method 3: Superimpositionon registrationpoint R,
with Bolton-nasionplanesparallel
4 Ghafari, Engel, and Laster
Fig. 4. Image of the digitized data for patient AD. (Original magnification x 1.5.)
.A
Fig. 5. Definition of the difference between methods 1 and 2
for the landmark A. This definition is true for every landmark
between any comparison of two methods. A = Landmark A
before treatment. A, = Landmark A after treatment as shown
with superimposition method 1. AZ = Landmark A after treat-
ment as shown with superimposition method 2. . . . = Dis-
tance between A, and A,.
- Method 4: Superimposition on Ba-N, registered
at CC
- Method 4a: Superimposition on Ba-N, registered
at N
The distribution of patients, tracings, and corre-
sponding number of superpositions are summarized in
Table I.
The data were collected with a Tektronix Plot 50
interactive digitizing system* at the Department of City
*Tektronix Inc., Beaverton, Ore
Am. 1. Orthod. Dent&c. Orthop.
‘?4ay 1987
and Regional Planning, Urban Simulation Laboratory,
University of Pennsylvania. The different methods of
superimposition were compared as follows:
Step I. On the initial pretreatment cephalomet~c
tracing, a vertical VL (obtained by using the “natural
head position”30) is drawn through nasion.
Step 2. The pretreatment tracing is placed on the
graphic tablet of the digitizing system over a millimeter
graded sheet The coordinate system for each patient is
so established that the y axis is parallel to the VL, the
x axis is perpendicular to the VL, nasion is placed on
a point of coordinates (x = 20 and y = lo), and the
whole of the tracing is situated in the quadrant where
both x and y coordinates are positive.
Step 3. The posttreatment tracing is then superim-
posed on the pretreatment tracing according to each of
the four methods described previously.
From each tracing and for each superimposition
method, nine landmarks were digitized before and after
treatment: PNS, ANS, A, B, pogonion, gonion, artic-
mare, sella, and nasion (Fig. 3). The x and y coordi-
nates of every digitized point were instantly and auto-
matically recorded on the data disk. For each patient,
4.5 x coordinates and 4.5 y coordinates of 45 pretreat-
ment points (9 identical points digitized 5 times), as
well as 45 x’ coordinates and 45 y’ coordinates of 45
VolumeP1 Cephalometric su~e~~mposi~ion 4
Number 5

II. Means and standard errors of differences between paired methods (1, 2, 3, 4, and 4a) for

MJN 0.9 i 0.2 0.7 + 0.0 1.0 -+ 0.1 1.2 -t 0.1 1.2 I 0.2 0.8 t 0.2
W/W 1.0 i- 0.1 0.7 -c 0.1 1.1 z!z 0.1 1.3 r 0.1 1.4 = 0.1 1.3 i 0.2
MO4 1.2 +- 0.2 0.8 ? 0.1 1.2 ST 0.2 1.4 + 0.2 1.4 t 0.2 1.1 2 0.1
W/M,, I.5 t 0.2 1.3 i 0.2 1.5 i 0.2 1.7 i 0.2 1.7 i 0.2 1.6 + 0.1
M*JM, 0.7 t 0.1 0.4 i 0.1 0.7 t 0.1 1.0 I 0.2 1.0 i 0.2 0.9 i 0.2
WM4 0.9 i 0.1 0.8 i: 0.1 0.9 2 0.1 1.0 r 0.1 1.0 i 0.1 1.0 r 0.1
WM4, 1.3 -t 0.2 1.3 2 0.2 1.4 2 0.2 1.4 r 0.3 1.4 i 0.3 1.5 + 0.2
M,IW 0.8 i- 0.2 0.7 2 0.1 0.8 i 0.2 1.0 ” 0.2 1.1 i- 0.2 1.1 4 0.3
wwa 1.4 + 0.3 1.3 i 0.2 1.4 k 0.2 1.5 i 0.3 1.5 2 0.3 1.7 i 0.3
M&L 0.9 r 0.1 1.0 * 0.1 0.9 -+ 0.1 1.0 i 0.i 1.0 f 0.1 1.0 t 0.1

No ‘clinically’-statistically significant differences observed under the hypothesis that a difference ~1 mm is clinically insignificant

Table III. Means and standard errors of differences between paired methods (1, 2, 3, 4, and 4a) for
cephalometric landmarks ANS, PNS, A, B, POG, and GON in boys (N = 13)

M,W 0.7 rt 0.1 0.6 f 0.1 0.8 + 0.1 0.9 + 0.2 1.0 It 0.2 0.8 + 0.1
M,!M, 1.0 * 0.2 0.7 r 0.1 0.8 k 0.2 1.3 + 0.2 1.4 c 0.3 1.2 + 0.2
N/M, 1.3 f 0.2 1.4 f 0.3 1.2 i 0.2 1.5 2 0.4 I.7 i- 0.4 1.6 + 0.5
MI/Mb 2.7 I!Z 0.3" 2.7 k 0.3" 2.8 k 0.3* 3.0 * 0.5" 3.2 + 0.5 3.1 i 0.5"
MZIM, 0.9 f 0.1 0.7 + 0.l 0.8 rt 0.1 1.1 -t 0.2 1.4 ” 0.2 1.1 i 0.2
N/M, 1.3 k 0.2 1.4 + 0.3 1.2 i 0.2 1.5 t 0.3 1.6 + 0.4 1.7 i 0.4
M*/W, 2.8 rt 0.3* 2.7 +- 0.3* 2.8 f 0.3* 3.0 i 0.4" 3.1 i 0.5 3.1 i 0.5"
K/M, 1.4 k 0.2 1.4 rt 0.3 1.4 t 0.2 1.7 i 0.3 1.9 + 0.4 1.7 + 0.4
M?M,, 2.8 rt 0.3* 2.7 f 0.4* 2.8 ri 0.4* 3.1 i 0.5* 3.3 2 0.5" 3.1 + o.s*
wma 2.1 * 0.3* 2.0 r 0.3 2.1 +- 0.3' 2.2 -f 0.3 2.! i 0.3 2.1 -L- 0.3

*‘Cli@ally’-statistically significant difference under the hypothesis that a difference CC1 mm is clinically insignificant (P < 0.01, jointly for
all tests).

posttreatment points (9 points per method) were re-
corded.
Fig. 4 represents the digitized image of the data
collected for the patient A.D. This image is seen on
the screen of the Tektronix 4052 and printed on the
plotter. The digitized data were stored in an IBM 4341
computer system and analyzed on an IBM 3101 com-
puter system. *
Before assessing differences between methods of
superimposition and in order to estimate the error of
measurement, three determinations per subject were
calculated on five subjects between method 1 and
method 2 for one landmark, PNS. An estimate of the
variance of these measurements that accounted for both
the variance between and within subjects was obtained:
or2 measurement = 0.33. This estimate indicated that
an average difference between the methods of 0.5 mm
could be detected with 80% power and the “type 1
*IBM Corporation, Danbury, Corm.
error” fixed at [Y = 0.05 (tt) with the use of 13 in-
dependent observations. Thus, the study sample con-
sisted of 13 subjects of each sex.
Differences in amount of change between the su-
perimposition methods were assessed independently for
each of the six following landmarks: PNS, ANS, A,
B, Pog, and Gon (Fig. 3). On each patient and for each
landmark, ten distances-the paired differences of five
posttreatment positions obtained by methods 1, 2, 3,
4, and 4a (Fig. 5)-were analyzed under the null hy-
pothesis that there was no difference between the paired
methods. Two methods were compared at a time. A t
test (single sample, based on the differences) examined
the average difference for each comparison. Compar-
isons with method 4a were valid mainly for changes in
the position of point A since Ricketts advocates method
4a only for that purpose.
RESULTS
All differences among all paired methods were sta-
tistically significant (P < 0.01). Because these results
 Ghafari, Engel, and Laster

sup=3rimpskm
,:? 1:3 1:4 1:4a 23 2:4 230 3:4 x4a 4:4G msthcdr

Fig. 6. Means and confidence intervals (95%) of differences between paired methods (1, 2, 3, 4, and
4a) for cephalometric landmark ANS in boys and girls (N = 26). *‘Clinically’-statistically significant
difference. P < 0.01, jointly for all tests.

Talsle IV. Means and standard errors of differences between paired methods (1, 2, 3, 4, and 4a) for
cephalometric landmarks ANS, PNS, A, B, POG, and GON in boys and girls (N = 26)

mm 0.8 i 0.1 0.6 rir 0.0 0.9 i. 0.1 1.0 +- 0.1 1.2 i 0.1 Q.8 I 0.1
WM 1.0 +- 0.1 0.7 i: 0.1 1.0 i 0.1 1.3 It 0.1 1.4 2 0.2 1.2 4 0.2
MN 1.2 -c 0.1 1.1 * 0.2 1.2 t 0.1 1.4 ” 0.2 1.6 -I 0.2 1.4 + 0.2
Ml/M,, 2.1 Ik 0.2* 2.0 i 0.2* 2.2 zt 0.2* 2.3 t 0.3* 2.4 -I- 0.3* 2.3 r!z 0.3”
Ma% 0.8 li: 0.1 0.6 + 0.1 0.8 +- 0.1 1.1 10.1 1.2 2 0.2 1.0 i: 0.2
M*lM, 1.1 r 0.1 1.1 ?I 0.2 1.0 * 0.1 1.2 k 0.2 1.3 +- 0.2 1.3 + 0.2
M*/W, 2.0 + 0.2* 2.0 L 0.2* 2.1 r 0.2* 2.2 i 0.3” 2.3 r 0.3* 2.3 Ir 5.3*
b&/M, 1.1 t 0.1 1.1 +- 0.2 1.1 k 0.1 1.3 I 0.2 1.5 + 0.2 1.4 + 0.2
MJM, 2.1 i 0.2* 2.0 -c 0.2* 2.1 c 0.3* 2.3 -+ 0.3” 2.4 t 0.4” 2.4 c 0.3”
Whll,* 1.5 s 0.2 1.5 -+ 0.2 1.5 t 0.2 1.6 i: 0.2 1.6 -c 0.2 1.5 L 0.2

*‘Clinically’-statistically significant difference under the hypothesis that a difference 51 m m is clinically insignificant (P < 0.01, jointly for
ail tests)

indicated that the difference between any two methods
of superimposition for any of the six landmarks con-
sidered was statistically significantly different from 0,
t tests were viewed under the hypothesis that a differ-
ence d I m m is insignificant clinically. The 1 m m stan-
dard for clinical significance was established on the
basis of t2a measurements to exclude with strong
confidence (95%) differences caused by variation in
measurement.
The results of comparisons based on clinically des-
ignated significance showed no statistically significant
differences between the methods for any of the six
landmarks in girls (Table II). In boys, clinically-statis-
tically significant differences showed only between
method 4a and each of methods 1, 2, and 3 for points
A, B, ANS, PSN, and Gon (P < 0.01, jointly for all
tests), between methods 4a and 3 for Pog (la < 0.01,
jointly for all tests), and between methods 4 and 4a for
points A and ANS (P < 0.01, jointly for alI tests)
(Table III). When the values for boys and girls were
pooled together, clinically-statistically significant dif-
ferences were found between method 4a and each of
methods 1, 2, and 3 for all six landmarks evaluated
(P < 0.01, jointly for all tests) (Table IV). Means and
confidence intervals (95%) of differences between
paired methods for each landmark in the total sample
(boys and girls) are presented in Figs. 6 through 11.
DISCUSSION
This study demonstrates that differences in inter-
pretation of facial change can be related to the method
of superimposition used. This conclusion is not sur-
Volume 91 Cephalometric ~~perirn~o~~t~o~
Number 5

mm PNS

Fig. 7. Means and confidence intervals (95%) of differences between paired methods (1, 2, 3, 4, and
4a) for cephalometric landmark PNS in boys and girls (N = 26). *‘Clinically’-statistically significant
difference. P < 0.01, jointly for all tests.

Fig. 8. Means and confidence intervals (95%) of differences between paired methods (1, 2, 3, 4, and
4a) for cephalometric landmark A in boys and girls (N = 26). *‘Clinically’-statistically significant differ-
ence. P < 0.01, jointly for all tests.

prising in view of the findings by Baumrind, Miller,
and Molthen3’ that errors associated with the act of
superimposition itself are sufficiently large to influ-
ence the interpretation of positional changes of facial
structures.
The question arises whether errors inherent to any
one of the methods compared are, if not equal, at least
similar in range. This question was not evaluated in the
present study. Baumrind, Miller, and Molthen com-
pared two methods of superimposition on cranial ref-
erences, which were used in this study: (a) superim-
position on SN, registered at S, and (b) superimposition
on optimal fit of anterior cranial base (best fii of the
anatomic structures of the floor of the anterior cranial
fossa, primarily the region between the anterior clinoid
processes and crista galli).3’ They concluded that ro-
tational errors for the best fit superimposition were
“unexpectedly somewhat smaller” than those for the
SN superposition and that errors of interpretation upon
maxillary or mandibular landmarks were “~erna~~ab~~
similar in character” between the two methods. More
recently, Pancherz and Hansen3’found the effect of the
registration error of the SN method upon the interpre-
tation of facial change to be less than the effect of
 Glzajari, Engel, and Laster Am. .I. Orthod. Dent&c. Orthop.
May 1987

..... .....

Fig. 9. Means and confidence intervals (95%) of differences between paired methods (1, 2, 3, 4, and
4a) for cephalometric landmark B in boys and girls (N = 26). “Clinically’-statistically significant differ-
ence. P < 0.01, jointly for all tests.

mm
PQG

T
i
I..... I . .. . . .. ..
1

Fig. 10. Means and confidence intervals (95%) of differences between paired methods (7, 2, 3, 4, and
4a) for cephalometric landmark POG in boys and girls (N = 26). *‘Clinically’-statistically significant
difference. P < 0.01, jointly for all tests.

registration enor of the best fit method. Differences in
the results of these investigations may be attributed to
methodology, including indentification of landmarks,
definition of anterior cranial base structures, film den-
sity and sharpness, and errors of tracing and mea-
surements.3’-34
Such errors are potentially inherent to any of the
methods evaluated in this study as each cephalometric
landmark exhibits a specific “envelope of error.“33 In
this respect, the differences evaluated reflect not only
differences in movement of landmarks, but also errors
in identification of these landmarks.
Although each of the methods evaluated is based
on a rationale recognizing the relative stability of the
cranial base, by using different parts of the cranial base
as references for superimposition, the various methods
interpret this concept of stability. The present study
demonstrated that differences among all paired methods
were statistically significantly larger than 0 (P < 0.01).
When viewed under the hypothesis that a difference s 1
m m was insignificant clinically, the results were dif-
ferent (Figs. 6 through 11). Although clinical signifi-
cance is probably an individual, not a standard, mea-
surement, the 1 m m standard for clinical significance
Volume 91 Cephalometrtc s~~e~~rn~osit~o~
Number 5

Fig. 11. Means and confidence intervals (95%) of differences between paired methods (1, 2, 3, 4, and
4a) for cephalometric landmark GON in bovs and girls (N = 26). *‘Clinically’-statistically significant
difference. P < 0.01, jointly for all tests.

was established on the basis of t 2a measurements to

exclude with strong confidence (95%) differences
caused by variation in measurement.
Nevertheless, significant differences were demon-
strated between method 4a and each of methods 1, 2,
and 3 for boys, and for the entire sample. Although
these differences were depicted for all six landmarks
evaluated, only differences in point A are of value to
the clinician because method 4a was solely advocated
to assesschanges in point A.” Thus, method 4a gives,
for the same person; an interpretation of anterior max-
illary change in position different from methods 1, 2,
and 3.
This result is proba’bly caused by the fact that
method 4a depicts the change in point A after sliding
basion-nasion plane upward and forward to register on
nasion. Subsequently, point A is translated downward
+
and backward relative to its posttreatment position de- Y
termined according to methods 1, 2, 3, and 4. This
Fig. 12. Average change in amount (vector length) and direction
movement is demonstrated in Fig. 12, which illustrates (vector angulation) observed for point A in the total sample
the average change in arnount (vector length) and di- (males and females) following superimpositions by methods 1,
rection (vector angulation) observed for point A.35 2,3,4, and 4a. Point A moves further downward and backward
Therefore, one may conclude according to method 4a with method 4a. (Original magnification x 20.)
that distalization of the maxillary complex occurred
following headgear therapy, when normal development 4 and 4a for point A in the entire sample may be at-
or simply a restraint of anterior maxillary growth would tributed to the occurrence of translational change of
be demonstrated by methlod 1, 2, 3, or even 4. Indeed, posttreatment point A after shifting Ba-N to register
when Ba-N is registered at point CC (method 4), no on N; a combination of rotational and translational
clinically-statistically significant differences with meth- changes3’accounts for the differences between method
ods I, 2, or 3 are revealed for point A (Fig. 8). The 4a and each of methods 1, 2, and 3. Also, small
lack of significance in the difference between methods amounts of forward and upward growth at nasion, there-
 Ghafari, Engel, and Laster
fore smaller shifts of Pa-N to register at N, may have
contributed to the lack of clinically-statistically signif-
icant difference between methods 4 and 4a in the entire
sample. However, differences between these methods
(4 versus 4a) in boys were significant for both points
A and ANS, suggesting the occurrence of signifi-
cant growth at nasion and/or significant translational
changes at A and ANS.
The fact that clinically-statistically significant dif-
ferences were found for boys and not girls was probably
due to timing of treatment in the population sampled,
and sex differences in absolute measurements and rate
of facial growth. Treatment of girls started at an average
age of 12.2 years, closer to or following the average
age of adolescent growth spurt in girls (12 years); treat-
ment of boys was initiated at an average age of 12.5
years, prior to the average adolescent growth spurt in
boys (14 years). Thus, relatively more growth should
have occurred in boys than in girls during the treatment
period evaluated. However, data on skeletal age of the
studied sample were not available to support this as-
sumption. Also, several reports suggest that during the
adolescent growth period, girls have smaller absolute
measurements and a slower rate of facial growth than
boyS.2,‘3,36
Pinally, there was no clinically-statistically signif-
icant difference registered for any of the mandibular
landmarks between any pair of the methods 1, 2, 3,
and 4, evaluated under the hypothesis that a difference
aI m m is insignificant clinically. If this hypothesis is
indeed valid, the various methods of registration on the
cranial base as applied in this study should yield similar
information about mandibular movement. This as-
sumption is made notwithstanding the fact that land-
marks do not represent the shape or relative position of
an entire curved body (mandible or maxilla),37 that spa-
tial changes are portrayed as linear displacements rather
than a generalized distortion,37 and the probability of
clinical significance being an individual measure. Such
propositions should temper the clinician’s determina-
tion of facial change based on cephalometric evaluation.
Although this study does not suggest a preference
among method.s 1, 2, 3, or 4, which use the cranial
base as a reference for cephalometric superimposition,
method 1 of best fit of the anterior cranial base may be
better than the others. Method 1 takes into consideration
the detailed individual anatomy of the cranial base,
rather than simplifying this anatomy into lines or planes
(SN, Ho-N, Ba-N) with different registration points (S,
R, CC), which may be modified by growth.
This assumption is supported by anatomic and his-
tologic studies of the cranial base.7 However, the ceph-
alometric tracings of the cranial base require reproduc-
Am. J. Orthod. Dent&c. Orthop.
May 1987
tion of as many structures of the base as possible, not
just tracing its general outline. Errors of registration
among operators were found to be “unexpectedly”
smaller than those for the SN superposition in a study
by Baumrind, Miller, and Molthen.3’ Another study by
Pancherz and Hansen3* reported opposite results. Pos-
sible reasons for these differences were discussed pre-
viously.
As for the validity of method 4az2regarding spatial
change of point A, it has yet to be proved because the
registration point N is itself subject to growth changes.
This study points clearly to potential differences in
interpretation between method 4a and the others. ore-
over, the results of this study warn against the com-
parison of orthodontic techniques on the basis of
different superimposition methads, and against the ex-
cessive reliance on one or another method of super-
imposition in interpreting growth andior orthodontic
treatment effects.
In conclusion, interpretation of facial change, es-
pecially in growing persons, should be made only in
reference to the superimposition method. As advocated
by many authors,31’38errors in using and int~~reting
cephalometric tracings may be reduced by replicating
the tracings, standardizing conditions to obtain good
quality films, and the use of ‘“rigorous and consen-
sually” accepted definitions of cepbalome~c land-
marks and tracing operations.” Although technical
advances are achieved with computerized assessment
of cephalometric radiographs, errors of inte~r~tation
are possible since human judgment remains a factor in
taking head films, and identifying landmarks and
structures.
The authors acknowledge Dr. Brainerd F. Swain for pro-
viding the material of this study and reviewing the manuscript,
Dr. Harvey Levitt for his helpful comments, and Drs. Stephen
H. Putnam and Hugh Miller for their cooperation in data
analysis.
REFERENCES
1. Moore WJ, Lavelle CLB. Growth of the facial skeleton in the
Hominoidea. New York: Academic Press, 1974.
2. Bambha JK. Longitudinal cephalometric roentgenographic study
of the face and cranium in relation to body height. J A m Dent
Assoc 1961;63:776-99.
3. Bjiirk A. The relationship of the jaws to the cranium. In: Lund-
Strom A, ed. Introduction to orthodontics. New York: McGraw
Hill Book Company, 1960:104-40.
4. Bjark A, Skieller V. Normal and abnormal growth of the man-
dible. A synthesis of longitudinal cephalomettic implant studies
over a period of 25 years. Eur J Orthod 1983;5:1-46.
5. De Caster L. The familial line: studied by a new line of reference.
Trans Cong (28th) Eur Orthod Sot 1952:50-5.
6. Ford EHR. Growth of the human cranial base. A M J ORTHOD
1958;44:498-506.
Volume 9 1
Number 5
7. Melsen B. The cranial base. Acta Odontol Stand 1974;32
(Suppl 62).
8. Brodie AG. On the growth pattern of the human head. Am J
Anat 1941;68:209-62.
9. Bjork A. The face in profile. An anthropologic x-rays investi-
gation of Swedish children and conscripts [Thesis]. Sven Tandlak
Tidskr 1947;40(5B).
10. Ortiz MH, Brodie AG. On the growth of the human head from
birth to the third month of life. Anat Ret 1949;103:31 l-33.
11. Brodie AG. Late growth changes in the human face. Angle
Orthod 1953;23:146-57.
12. Steiner CC. Cephalometrics for you and me. AM J ORTHOD 1953;
39:729-55
13. Nanda RS. The rates of growth of several facial components
measured from serial cepbalometric roentgenograms. AM J
ORTHOD 1955;41:658-73.
14. Bjork A. Cranial base development. A follow up x-ray study of
the individual variation in growth occurring between the ages
12 and 20 years, and its relation to brain case and face devel-
opment. AM J ORTHOD 1955;41:198-225.
15. Krogman WT, Sassouni V. Syllabus in roentgenographic cepha-
lometty. Philadelphia: College Offset Press, 1957:226-32.
16. Fishman LS. A longitudinal cephalometric study of the normal
craniofacial profile, utilizing a proportional analysis of skeletal,
soft tissue, and dental structures. Int Dent J 1969;19:351-79.
17. Wei SHY. The variability of roentgenographic cephalometric
lines of reference. Angle Orthod 1968;38:74-8.
18. Stramrud L. External and internal cranial base. A cross-sectional
study of growth and of association in form Acta Odontol Stand
1959;A:239-66.
19. Ricketts RM, Schulhof RJ, Bagha L. Orientation sella-nasion or
Frankfort horizontal. AM J ORTHOD 1976;69:648-54.
20. Broadbent BH. Measurement of dentofacial changes in relation
to the cranium. In: Anderson D, ed. Practical orthodontia.
5th ed. St. Louis: The C V Mosby Company, 1935:184-204.
21. Broadbent BH. The face of the normal child. Angle Orthod 1937;
7:183-208.
22. Ricketts RM. The bioprogressive therapy. Denver: Rocky Moun-
tain/Orthodontics, 1979.
23. Ricketts RM, et al. A foundation for cephalometric communi-
cation. Ahl .I ORTHOD 1960;46:330-57.
24. Delattre A. Du crane animal au crane humain. Paris: Masson,
1951.
25. Delattre A, Fenart R. Developement du c&e. Bulletin et MCm-
oires de la SocittC Anthropologique. Paris: 1955;6:159-73.
Cephalometric sL~~eri~pos~tion
26. Delattre A, Fenart R. L’hominisation du c&e etudiee par la
methode vestibulaire. Paris: Ed du Cent of Nat Recherche Scient,
1960.
27. Fenart R. Le Vestibulographe. Bulletin et MCmoires de la So&%
Anthropologique. Paris: 1969;12:363-70.
28. Cousin RP. Etude en projection sagittale de cranes d’enfants
orient& dans les axes vestibulaires [Doctoral thesis]. University
of Paris, 1969.
29. Cousin RP, Fenart R. Les profils squelettiques crania-faciaux
chez I’homme. Arch Anat Path01 1970;18:285-92.
30. Moorrees CFA, Kean MR. Natural head position, a basic con-
sideration for the analysis of cephalometric radiographs. Tr. Eur
Orthod Sot 1958;34:68-86.
31. Baumrind S, Miller D, Molthen R. The reliability of head film
measurements. 3. Tracing superimposition. AM J ORTHOO1976;
70:617-44.
32. Pancherz H, Hansen K. The nasion-sella reference line in ceph-
alometry: A methodologic study, AM J ORTHOD 1984;86:
427-34.
33. Baumrind S: Frantz R. The reliability of bead film measure-
ments. 1. Landmark identification. AM J ORWOD 19?1;60:
111-27.
34. Baumrind S, Fran@ R. The reliability of head film measure-
ments. 2. Conventional angular and linear measures. Ahl .I
ORTHOD 1971;60:505-17.
35. Engel FE, Ghafari J. Evaluation of amount and direction of
spatial change illustrated by cephalometric superimpositions [In
preparation].
36. Hunter CJ. The correlation of facial growth with body height
and skeletal maturation at adolescence. Angle Orthod 1966;36:
44-54.
37. Moyers RE, Bookstein FL. The inappropriateness of conven-
tional cephalometrics. AM J ORTHOD 1979;75:599-617.
38. Houston WJB. The analysis of errors in orthodontic measure-
ments. AM J ORTHOD 1983;83:382-90.
Reprint requeststo:
Dr. Joseph Ghafari
Department of Orthodontics
School of Dental Medicine
University of Pennsylvania
4001 Spruce Street, Al
Philadelphia, PA 19104