How accurate is model planning A. Sharifi, R. Jones, A. Ayoub,K. Moos, F. Walker, B. Khambay,

er-impacted anteriorly than predicted by

supine or anaesthetized patient would appear to be the principal reasons for errors.Keywords: model surgery; orthognathic sur-gery; articulator; osteotomy.

Current methods of orthognathic surgical

ware.6,13

Model surgery depends on the accuracy

tion of the occlusal plane to the Frankfort

they found a mean difference of 58, which

One of the commonest errors in orthog-

accompanied by inherent errors in the

lator.4,11

Int. J. Oral Maxillofac. Surg. 2008; 37: 10891093://deformity and malocclusion. Analysis ofModel surgery planning on dental castsis used for the final correction of facial

horizontal on the Hanau articulator andcompared this with lateral cephalograms;

adjustment of the instrument to thepatients face and transfer to the articu-planning involve clinical evaluation,photographs, freehand surgical simula-tion based on cephalometric tracingand then transferred to study model sur-gery, and computerized prediction soft-

of the recording of the dental occlusion inthe retruded position and the face bowtransfer to the articulator. These record-ings both have inherent inaccuracies.BAILY and NOWLIN1 measured the angula-

nathic model surgery occurs in mountingthe models on the articulator. The accu-racy of the face bow transfer may differfrom one type of face bow to another. Thekinematic face bow transfer record isInaccuracies are associated with the transfer of prediction planning to model surgeryplanning and prediction, which should be eliminated to improve the accuracy andpredictability of orthognathic surgery.model surgery. The amount of mandibular setback was more than that predicted bymodel surgery. None of the differences between prediction planning and actualsurgical changes was statistically significant at p < 0.05. Inaccuracy with the facebow recording, the intermediate wafer, and auto-rotation of the mandible in thefrom model surgery prediction and a statmaxilla was more under-advanced and ovthe model surgery allows the transfer ofplanned three-dimensional movements forthe surgical correction of complex dento-facial deformities.8

0901-5027/1201089+ 05 $30.00/0 # 2008 Interncorresponded to a 70% error. The three-dimensional accuracy of the position ofthe upper first molar was highly variableusing four different Hanau facebows.16

ational Association of Oral and Maxillofacial SurgeoAccepted for publication 13 June 2008Available online 28 August 2008for orthognathic surgery?A. Sharifi, R. Jones, A. Ayoub, K. Moos, F. Walker, B. Khambay, S. McHugh: Howaccurate is model planning for orthognathic surgery?. Int. J. Oral Maxillofac. Surg.2008; 37: 10891093. # 2008 International Association of Oral and MaxillofacialSurgeons. Published by Elsevier Ltd. All rights reserved.

Abstract. The purpose of this study was to evaluate the accuracy of model surgeryprediction after orthognathic surgery and to identify possible errors associated withthe prediction process. The study included 46 patients who had undergoneorthognathic surgical procedures; 22 in Group Awho had had a Le Fort I osteotomy;and 24 in Group B who had had a Le Fort I osteotomy and mandibular setbacksurgery. The immediate postoperative and preoperative lateral cephalograms wereanalysed to calculate surgical changes; these were compared with those obtained

istical analysis was undertaken. TheS. McHughGlasgow Dental Hospital & School, GlasgowUniversity, 378 Sauchiehall Street, Glasgow,G2 3JZ, UKdoi:10.1016/j.ijom.2008.06.011, available online at http www.sciencedirect.comClinical PaperOrthognathic SurgeryAssessing the accuracy of orthognathicsurgery by comparing surgical changeswith model surgery is prudent. It is impor-tant to recognize why the pre-planned

ns. Published by Elsevier Ltd. All rights reserved.

and those predicted. SPSS version 11.5

ior point of the posterior plate. The pro-

model. In 36% (n = 8) the maxilla wasunder-advanced and in 14% (n = 3) itwas over-advanced in comparison withprediction planning (Fig. 1). Table 1 shows

There was no statistically significantdifference between the two groups regard-ing the magnitude of disparities of actual

1090 Sharifi et al.

th.

Table 1. Differences between the actual changes and those predicted for movements in Group Aalong the X co-ordinate; a negative sign means less advancement surgically than that predictedby model surgery

Mean SD Median IQ (25%:75%)

ANS 1.16 1.98 0.80 2.97:0.24PNS 0.20 2.06 0.47 1.04:1.44A 0.20 2.21 0.10 2.36:1.40UIE 0.14 2.34 0.25 2.72:1.08UME 0.32 2.36 0.21 2.33:1.40P point 0.86 2.33 1.03 2.85:0.66M & M 0.34 2.15 0.21 2.49:0.86Abbreviations: A: Maxillary A Point; ANS: Anterior Nasal Spine; IQ: Interquartile; M&M:Mean of all the maxillary landmarks; P: The maximum concavity of palate; PNS: PosteriorNasal Spine; SD: Standard Deviation; UIE: Upper Incisor Edge; UME: Upper Molar Edge.(Statistical Package for Social Science)was used to analyse the data.model surgery is not always reflected inthe final outcome.

Materials and methods

The most common group of patients trea-ted in the authors unit are those withmaxillary hypoplasia and those with ClassIII skeletal deformity due to maxillarydeficiency and mandibular prognathism.This investigation was carried out on 46

patients: 22 in Group A who had had a LeFort I maxillary advancement and impac-tion; and 24 in Group B who had hadbimaxillary osteotomies, maxillaryadvancement and impaction with a man-dibular setback. None of the cases hadsegmental surgery or facial asymmetry.In all cases the maxillary position wastransferred to the articulator using an aver-age value condylar face bow (Dentatusface bow). The upper and lower casts weremounted on a Dentatus semi-adjustablearticulator.The availability of preoperative and

immediate postoperative lateral cephalo-grams and casts was mandatory for thisinvestigation. The following landmarkswere digitized on all the pre- and post-operative radiographs: ANS (anteriornasal spine), PNS (posterior nasal spine),A point, UIE (upper incisor edge), UME(upper molar edge), P (the most concavepoint of the palate), B point, LIE (lowerincisor edge), LME (lower molar edge),Pg (pogonion), Me (menton), Ge (genion:the most posterior point on the lingualcortical aspect of the symphysis abovethe genial tubercle and usually 12 mmbelow the level of the lower incisor apex).The pre- and postoperative lateral

cephalograms were superimposed on thebest fit of the anterior cranial base8 usingAdobe Photoshop 7.0 (Adobe incorpo-rated 2004 USA).The Frankfort horizontal (FHP) was

chosen as the reference line for this study(X axis). A perpendicular line to the FHPwas drawn from the centre of the sella (Yaxis). All pre- and postoperative land-marks were measured to extract X andY coordinates. The actual changes follow-ing surgery were measured at A and Ppoints for the Le Fort I osteotomy and at Band Ge points for mandibular setbacksurgery. These changes were comparedwith the changes predicted by the modelsurgery. Students t-test was used to iden-tify the skeletal significance of the differ-ence between the actual surgical changescedure was repeated after 1 month. TheDahlberg formula was applied to assessthe magnitude of errors.5,10

Results

The errors in the landmark locationstraced from the cephalogram were only0.2 mm across the X axis and 0.3 mmacross the Y axis.In Group A, in 50% of the cases, the

actual forward movement of the maxillawas accurate to within 1 standard deviationof the mean, with that of the predictionError of methods

The reproducibility of the superimpositionof the lateral cephalographs was assessed.The pre- and postoperative radiographs of5 randomly selected cases were superim-posed on the cranial base of the skull, andthe following 4 landmarks were placed:upper incisor edge, lower incisor edge,lower antero-inferior point of anteriorfixation plate, and the most postero-infer-

Fig. 1. Difference in advancement between(maximal concavity of the palate) in Group Athe tendency toward under-advancement ofthe maxilla.Fig. 2 shows the accuracy of the pre-

diction planning in Group A at point P.Similar to the horizontal changes, about50% (n = 11) of the cases had anteriormaxillary impaction similar to the predic-tion values (within 1 standard deviation ofthe mean). In about 33% (n = 7) the max-illa was over-impacted, and it was under-impacted in the remainder (n = 4)(Table 2). Posterior maxillary impactionat PNS was similar to model surgery pre-diction in 64% (n = 14) (Table 2).In Group B (n = 24), a similar pattern of

differences between the actual surgicalchanges and the prediction movementwas also identified (Tables 3 and 4). Accu-rate maxillary advancement as predictedwas achieved in 46% of cases (n = 11),with 37% (n = 9) showing under-advance-ment and 17% (n = 4) over-advancement(Fig. 3). Anterior over-impaction wasnoted in 33% (n = 8) and posteriorunder-impaction in 25% (n = 6) of cases(Figs. 4 and 5, Table 4).

e actual changes and prediction at point P

How accurate is model planning for orthognathic surgery? 1091

tun

tuTable 2. The differences in mm between the acin Group A along the Y axis; a negative sig

Fig. 2. Difference in impaction between the acconcavity of the palate) in Group A.maxillary surgical movement and the pre-dicted values from orthognathic models.In Group B, 46% (n = 11) had an accuratemandibular setback as predicted in thestudy models. Over-correction wasdetected in 37% (n = 9) and under-correc-tion in 17% (n = 4) (Fig. 6). The meanvalues showed a tendency towards settingthe mandible more posteriorly than pre-dicted.None of the differences detected in

Groups A and B between the actual sur-gical movements and the predictions from

predicted by model surgery

Mean SD

ANS 0.29 2.38A 0.37 2.97UIE 0.50 2.13P point 0.94 2.01

Table 3. The differences between the actualadvancement in Group B along the X axis; a nethan that predicted by model surgery

Mean SD

ANS 0.37 2.20PNS 0.73 2.10A 0.50 2.68UIE 0.20 1.98UME 0.39 2.77P point 0.63 2.92M & M 0.06 2.1

Table 4. The differences between the actualimpaction in Group B along the Y axis; a nega

Mean SD

ANS 0.42 1.85A 0.25 2.30UIE 0.06 2.45P point 0.65 2.09al changes and those predicted for movementsmeans more impaction surgically than that

al changes and prediction at point P (maximalstudy models was statistically significant(p < 0.05). The difference at ANS wasclose to statistical significance atp = 0.051.

Discussion

In this study, the maxilla showed a ten-dency to under-advancement comparedwith the predicted movement in about33% of the cases in both groups. Thiscould be a result of the inaccuracy oftransferring the maxillary plane angle

Median IQ (25%:75%)

0.28 1.96:1.350.83 1.69:1.871.13 2.42:1.101.16 2.5:0.22

changes and those predicted for maxillarygative sign means less surgical advancement

Median IQ (25%:75%)

0.05 1.94:1.190.21 2.41:-0.210.22 1.79:2.630.47 1.13:1.440.62 1.87:2.381.17 2.68:1.160.06 1.67:1.07

changes and those predicted for maxillarytive value means over impaction

Median IQ (25%:75%)

0.10 1.90:0.810.31 1.94:0.970.14 1.64:0.490.78 2.58:0.57and the inaccuracy of recording the man-dibular position.The semi-adjustable articulator used for

model surgery in this study was originallycreated for prosthetic dentistry. Its facebow was designed to transfer the relation-ship of the maxilla to the terminal hingeaxis of the mandible. To accomplish this,the posterior end of the face bow is alignedto the terminal hinge axis (middle of con-dyle), and the anterior end is aligned to theorbitale. These points define a plane calledthe axis-orbital plane,7,9 which is relatedto the upper cross arm of the articulator tomount the maxillary occlusal model.The upper cross member of the articu-

lator represents the Frankfort horizontalplane, which may not be horizontal(Fig. 6), as assumed in designing the exist-ing semi-adjustable articulator. This couldcause errors in model planning becausemost vertical measurements made duringmodel surgery are parallel to a referenceline, usually the upper arm of the articu-lator. When the maxilla is impacted in ahorizontal plane superiorly on the articu-lator, the magnitude of the upward move-ment is related to a horizontal referenceline. The extent of the impaction is dif-ferent when the cast has been mountedwith an angle between the occlusal planeand the upper member of the articulator,which is different from the real anglebetween the occlusal plane and the Frank-fort horizontal in that patient, for which adifferent amount of impaction would berequired.7 A more anterior maxillaryimpaction would be needed surgicallythan that predicted by model surgery tocompensate for the steepness of the max-illary plane angle on the articulator; thiswas true in 33% of the cases in this study.When using a conventional articulator

for orthognathic surgery, it is essential thatthe angle between the occlusal plane andthe Frankfort horizontal for the patient isthe same as the angle between the occlusalplane and the upper member of the articu-lator on the maxillary model. If this isincorrect, the result of the model surgeryis erroneous.The authors suggest that every indivi-

dual lateral cephalogram should bechecked for the accuracy of the mountingof the maxillary cast on the articulator.Further research is needed to prove andconfirm the accuracy of this hypothesis.GATENO et al.9 showed that if there were

a 128 difference between the Frankforthorizontal plane and the axis-orbital plane,with bimaxillary model surgery, the max-illa would move 15% less than desired formaxillary advancement. In the presentstudy, in about 33% of cases, the maxilla

1092 Sharifi et al.

Fig. 5. Difference in impaction between the actual changes and predictions at PNS in Group B.

Fig. 4. Difference in impaction between the actual changes and predictions at point P (maximalconcavity of the palate) in Group B.

Fig. 3. Difference in advancement between the actual changes and predictions at point P(maximal concavity of the palate) in Group B.showed under-advancement from that pre-dicted for both groups.The other source of error is the differ-

ence in the patients mandibular positionwhen supine and upright; the mandibletends to be positioned more posteriorlywhen the patient is lying down and themouth has been actively closed into therelaxed position of centric occlusion.12

BOUCHER and JACOBY3 reported that themandible was positioned up to 2 mmmoreposteriorly in the anaesthetized patient. Aslight posterior displacement, with a ver-tical drop of the condyle (mean 2.4 mm)under general anaesthesia was reported.15

BAMBER et al2 recommended recording thecentric relationship in the supine con-scious position when planning bimaxillaryosteotomies. All the cases in the presentstudy had the wax bite and face bow takenin the upright position and in centricocclusion. This would have registeredthe mandible in a more anterior positionthan its location when the patient waslying flat under general anaesthesia. Oncethe maxilla was down-fractured at a LeFort I level, the mandible would be theonly reference point to adjust the maxil-lary position in an antero-posterior direc-tion using the occlusal wafer. Therefore,less maxillary advancement would beachieved than predicted on the articulatorowing to the more posterior mandibularposition. The mandible has been over-corrected (more setback) to compensatefor maxillary under-advancement.The other possible source of errors in

planning orthognathic surgery is inherentin the nature of mandibular hinge axis andthe consequences of its application inprediction planning. The true hinge axisof mandibular rotation is the axis exactlylocated as a result of the combined rotationand translation of the condyles. It has beenshown that an insignificant gliding of thecondyles of 0.5 mm significantly displacesthe located hinge axis. The inaccuracy inregistering and transferring the true hingeaxis of the condyle to the articulator wouldcause errors in predicting mandibular rota-tion secondary to maxillary surgery. Usingthe current instrumentation, discrepanciesbetween the true hinge axis and the trans-ferred axis for the mounted casts on semi-adjustable articulators are inevitable.14

Intermediate wafers are another majorsource of inaccuracy. After a Le Fort Iosteotomy and mobilization of the max-illa, the wafer places the maxilla in thepre-planned position relating it to themandible for fixation. Under generalanaesthesia, the muscles of masticationare relaxed and the mandible wouldnot serve as a fixed reference plane for

5. DahlbergG. Statistical method for med-ical and biological students. Interscience

How accurate is model planning for orthognathic surgery? 1093

na 1maxillary surgery. In addition, the wafercannot be placed exactly in the pre-planned position because there are noanatomical references points outside theosteotomy cuts against which a check canbe made. An instrument that would placethe intermediate wafer in its correct rela-tion to the Frankfort horizontal or visualaxis plane should be designed. Thisinstrument should be able to transfer reli-able anatomical references (e.g. externalauditory meatus and nasion) from patientto articulator. If there were clear anato-mical references, which act as externalframes of reference, these could guide thewafer into the planned position moreaccurately.There were no statistically significant

differences between the predicted andactual changes, mainly due to the smallsample size, however, this study has iden-tified clear differences in some casesbetween what was planned and achieved.These discrepancies may produce an unac-ceptable occlusion, therefore every effortshould be made to address the disparity

Fig. 6. Frankfort plane. (After DOWNS WB. Abetween model planning and surgicalplanning.This article did not investigate the clin-

ical significance of the mismatch betweenmodel prediction and actual surgicalchange, but it did prove its existenceand highlight ways of improving orthog-nathic model surgery. The use of a spe-cially designed orthognathic articulatorthat takes into consideration anatomicalvariations, accurately locating the mand-ible in three dimensions and reproducingits autorotation, should improve orthog-nathic model surgery.New York 1940.6. Donatsky O, Hillerup S, Jorgensen

JB, Jacobsen PU. Computerised cepha-lometric orthognathic surgical simula-tion, prediction and postoperativeAcknowledgements. The authors wouldlike to thank Mrs A Maguire and Ms AMcCormack for their help in compilingthis manuscript.

References

1. Bailey JO, Nowlin TP. Accuracy ofFrankfort plane transfer of the Hanauarticulator. J Dent Res 1981: 60: 531.

2. Bamber MA, Firouzai R, Harris M,Linney AD. A comparative study oftwo arbitrary face bow transfer systemsfor orthognathic surgery planning. Int JOral Maxillofac Surg 1996: 25: 339343.

3. Boucher L, Jacoby J. Posterior bordermovement of the human mandible move-ments. J Pros Dent 1961: 11: 836.

4. Bowley JF,MichaelsGC, Lai TW, LinPP. Reliability of face bow transfer pro-cedure. J Pros Dent 1992: 67: 491498.

lysis of the dentofacial profile. Angle Orthodevaluation of precision. Int J Oral Max-illofac Surg 1992: 21: 199203.

7. Ellis E, Tharanon W, Gambrell K.Accuracy of face bow transfer: Effects onsurgical prediction and post surgicalresult. J Oral Maxillofac Surg 1992: 50:562567.

8. FONSECA RJ. Oral Maxillofac Surg. Saun-ders: Philadelphia, 2nd Vol. 2000, Chap-ter 5, p. 98.

9. Gateno J, Forrest KK, Camp B. Acomparison of three methods of facebow transfer recording: Implication fororthognathic surgery. J Oral MaxillofacSurg 2001: 59: 635640.10. Ghafari J, Engel FE, Laster LL.Cephalometric superimposition on thecranial base: A review and a comparisonof four methods. Am J Orthod DentofacOrthopod 1987: 91: 403413.

11. Gold BR, Setchell DJ. An investiga-tion of the reproducibility of face bowtransfer. J Oral Rehab 1983: 10: 495503.

12. Helkimo M, Ingervall B, CarlssonGE. Variation of retruded and muscularposition of mandible under recordingconditions. Acta Odonto Scand 1971:29: 423427.

13. Loh S, Yow M. Computer prediction ofhard tissue profile in orthognathic sur-gery. Int J Adult Ortho Orthognat Surg2002: 17: 342347.

14. Lotzmann U. Considerations of preci-sion and consistence of mandibular trans-verse hinge axis Zentrum fur Zahn-Mund-und Kieferheilkunde, 99. derGeorg-August-Universitat Gottingen1990: p. 372379.

15. McMillan LB. Border movements ofhuman mandible. J Pros Dent 1972: 27:524.

16. OMalley MA, Milosevic A. Compar-ison of three face bow/semi adjustable

956: 26: 192212.)articulator systems for planning orthog-nathic surgery. Br J Oral Maxillofac Surg2000: 38: 185190.

Address:Ashraf F. AyoubGlasgow Dental Hospital & SchoolProfessor of Oral & Maxillofacial Surgery378 Sauchiehall StreetGlasgowG2 3JZUnited KingdomTel: +44 141 211 9604Fax: +44 141 211 9601E-mail: a.ayoub@dental.gla.ac.uk

How accurate is model planning for orthognathic surgery?Materials and methodsError of methodsResultsDiscussionAcknowledgementsReferences