J. Dent. Research 55:B295, 1976.

A Longitudinal Evaluation of the Burlington Growth Centre Data
Gordon W. Thompson and
Frank Popovich
Division of Clinical Sciences, University of Toronto
Faculty od Dentistry, Toronto, Ontario- Canada

In 1952, The Faculty of Dentistry at the University of Toronto established a Growth Centre in Burlington, Ontario for research into preventive and interceptive orthodontics. Burlington was chosen because it was close to the University of Toronto and was a typical small residential town. ln 1952, it had a population of approximately 9,000 but by 1972 it was 90,000.

Material and Methods
The Burlington Growth Centre data consist of six study groups (Table 1) plus 307 parents and 111 siblings. The 1,258 children in the Burlington sample represented 85 to 90% of the children in Burlington within the specified ages at the time the study was started in 1952. Therefore, the initial sample was representative, rather than selective, of a modern Canadian suburban community. ln the Serial Experimental sample, approximately 20% dropped out by 12 years of age. Comparisons were made at the early age-levels between the children who remained in the sample and those who ceased going to the clinic but did not reveal any significant differences between these two groups in the craniofacial size, shape or maturity level.

The children in the Serial Experimental sample were examined and records were taken from age 3 to 21 years, and most of these individuals had reached 20 years of age by August, 1971. The following records were made during the initial and annual visits to the Centre: case history; data compiled from an oral examination which included periodontal evaluation; six cephalometric radiograms employing high-kilovoltage technique, one wrist radiogram for determining the carpal index; impressions and wax bite for dental casts; intraoral radiograms where necessary; height and weight records; anthropometric measurements, and social histories and electromyographic records for some subjects.

The collection of the Burlington Growth Centre data was completed in 1971. The raw consists of more than 48,OOO cephalograms, 8,000 hand and wrist radiographs, 8,000 medical and dental histories, and 8,000 sets of dental diagnostic models.

In order to maintain the quality of the records, the cephalograms and dental casts have been duplicated. The originals are stored in cabinets and the duplicates are used for the various investigations.

To facilitate the research program, a computerized data bank was established to provide convenient storage, immediate accessibility, file updates and additions, and data editing. The cephalometric and dental cast data were digitized by using the Growth Centre's Gradicon 100 digitizer. This unit is used to derive the Cartesian co-ordinates of a point. These dimensions are then converted into measurements by the Facultv's IBM 1130 computer and the University's IBM 370-165 computer.

For growth analyses, the data were corrected for enlargement and distortion. The measurement criteria for each project is based on the measurement objectives of the Burlington Growth Centre; that is to derive reproductive measurements with minimum errors. To this end, the measurement error and reliability were derived from analysis of variance components. Subjective data are very seldom entered in the data bank. In this way, the measurements and data analyses of individual projects can be accumulated in the data bank so that other investigators or the staff of the Burlington Growth Centre can use them at a later date.

The treatment status of each individual was carefully categorized. The treated children were usually compared with the nontreated sample, and they were included in any analysis when there was no significant difference. Of the total sample, 30.2% received some orthodontic treatment (Table I).

Treatment status of the children in the 
Burlington Growth Samples
Sample size
Treated %
Serial Experimental
207 (66.3)
Serial Control (C-6)
59 (20.0)
Control Age 8 (C-8)
40 (18.2)
Control Age 10 (C-10)
29 (13.4)
Control Age 12 (C-12)
45 (20.9)
380 (30.2)

The Burlington Growth Centre was initiated 25 years ago. For the first 20 years, the emphasis was on data collection because limited funds and time were available for data analyses. During the last five years, the emphasis has been on the creation of the conputerized data bank, and longitudinal analyses. It is our intention to outline briefly some of the research projects, which have been or are presently being conducted in the fields of epidemiology, craniofacial growth, and orthodontic treatment. This longitudinal overview describes the projects and the types of analyses employed in these studies.


EPIDEMIOLOGY - The case histories, cephalograms, and dental casts of the total sample were examined for the degree of severity or absence of a maxillary diastema, missing teeth, sucking habits, and malocclusion status. The diastema analysis was based on the 471 cases that had complete records at ages 9 and 16 years. Of the 471 children who were evaluated, 230 (48.8%) had a diastema of 0.5 mm or greater at age 9 years. However, at age 16 years, only 44 (9.4%) had a diastema.

After comparing several characteristics associated with the persistence of diastema at age 16 years, generalized spacing was found to be the most significant factor. An analysis of the frenum type revealed that the children who had a space greater than 0.5 mm at age 16 years had a higher prevalence of frenae with low thin and low thick attachments. Children with a persisting diastema had intermaxillary sutures with spadeshaped or W-shaped bone. These suture types were not found in children with crowding.

Of the diastemas that closed, the closure took place primarily between the ages of 9 and 12 and was completed by 16 years. In those children whose diastemas did not close completely by age 16, most of the closure that did occur was also between 9 and 12 years age.

There was a definite association between age and occlusion type in the total sample of 1,258 children. The Class II malocclusion rate increased from 21.5% at the ages of 3 and 4 to 41.9% at the age of 12 years. The Class I cases remained close to the average of 44.0% at all ages from 3 to 21 years, while the normal occlusion rate decreased with age. Thumb sucking was associated with a significant increase in the prevalence of malocclusion. When the children stopped sucking before the age of 6, the percentage of Class II malocclusions was still higher than among those in the nonhabit group. However, in older children (those between 8 and 12 years of age) ,there was no association between Class II malocclusion and a habit that had stopped. This lack of an association could be due to the increased time interval during which there was a greater chance for self-correction after the habit had stopped. The children in the 6 year-old group would probably have demonstrated a similar prevalence after they too had time for self-correction.

The Class II malocclusion rate among children in the Serial Experimental sample increased with an increase in the age at which the habit had stopped; that is, the malocclusion rate of those who stopped after age 6 was higher than that of children who stopped between the ages of 3 and 6 years. In turn, the latter group had a higher rate than the children who stopped before the age of 3 years.

From the total sample, 1.191 children were studied for congenitally missing teeth. Of this group, 7.4% had missing teeth. When considering both sexes combined, most frequently only one or two teeth were congenitally missing. The probability of a tooth missing was about the same in both sexes but more males were missing just one tooth and more females were missing two or more teeth.

The probability of a specific tooth being missing was small; for example, the most frequently missing tooth, the mandibular second premolar, was absent in only 2.1% of the cases. Conversely, the probability of an individual having one or more missing teeth was 7.4%. If one tooth was missing the probability was greater that other teeth were also missing. As well, the probability was greater that a tooth would be missing bilaterally, if it was absent on one side.

Of the 521 children who had at least two records taken between the ages of 3 and 21 years and at least one record around 16 years, 22.3% had at least one third molar missing. The lack of significant sex difference in the congenital absence of third molars suggests that the probability of congenitally missing third molars is not sex dependent. From probability studies with both sexes combined, it can be generalized that when third molars are absent, there is a greater chance that other teeth will also be missing.

Third molars were more frequently missing in the mandible than in the maxilla. This greater decrease in number of teeth in the mandible is in accord with an evolutionary trend to greater reduction in the lower face than in the mid-face region. The probability of other teeth being missing was greater if the third molar was absent in the mandible than in the maxilla.

CRANIOFACIAL GROWTH ANALYSES - Based on the longitudinal growth direction of the chin relative to sella-nasion from age 4 to 17 years, 120 males and 90 females of the Serial Experimental sample were categorized into sex-specific percentiles. Equal-sized groups of vertical, horizontal-vertical, and horizontal growth pattern cases of each Sex were used-30 in each female group and 40 in each of the male groups. The analyses were based on the comparison between the vertical and horizontal groups. The average dimensions of these groups were compared by means of Studentís test to determine whether there were any statistically significant in size. A total of 61 lateral cephalometric measurements were compared.

Correlation of the growth direction of gnathion and other landmarks from age 14 to 17 years in the 23 serial experimental cases
Key Ridge
A pt
B pt

Most dimensions that were significantly different between the horizontal and vertical groups in one sex, were also siginifican in the other. The position of a landmark was more often different as a result of the horizontal rather than the vertical component. With few exceptions, the mean differences between the vertical and horizontal groups became significant earlier in the females than in the males. Some of these age differences were as great as 9 years. The horizontal growers were characterized by larger horizontal growth components, with a larger maxilla, larger mandibular ramus and body, a smaller anterior face height, but not a larger overall face-size (Fig1).

Fig. 1 - Facial areas that were larger in the horizontal or vertical group of males and females. Arrows demonstrate dimensions that were larger in the horizontal group and area that was smaller in the horizontal group and larger in the vertical group.

The growth directions from age 4 to 17 years of the 30 vertical and 30 horizontal female growers were compared (Fig 2) . The growth directions of the landmarks were significantly different between the two groups. When the two sets of growth directions were superimposed on sella with a 22 degree rotation, it is readily apparent that the growth directions of a particular landmark tended to be parallel. That is, once the adjustment of 22 degrees had been made in the cranial base the directions of growth were similar in the two groups.

To clarify further the inter-relation-ships of growth in the face,correlations were derived for the growth directions of the total Serial Experimental sample of 233 who had completed records. All of the correlations of the growth directions were statistically significant. Gnathion at the lower part of the chin was highly related to other mandibular growth directions, moderately related to the maxillary growth directions, and significantly related to the posterior cranium growth directions (Table 2).

Fig 2- The average age 4- and 17-year outlines of the 30 horizontal growers superimposed on the age 4- and 17-year outlines of the 30 vertical growing cases with a 22-degree rotation of the sella-nasion line. Growth direction of an anatomical landmark in the two growth direction groups tended to be parallel since the correction for the growth direction differential had been made.

The lateral and posteroanterior cephalograms of this group of 120 males and 90 females were plotted out at the enlarged size as as growth-type specific templates. As noted earlied, there were three lateral growth direction groups. The lateral templates can be used to determine which growth pattern an individual approximates most closely for that particular age and sex. All anteroposterior measurements are made parallel to the mean occlusal line. All vertical measurements are made perpendicular to this line. The mean occlusal line is used because it is in the dental area and it more closely approximates the natural growth axis. This enables small amounts of growth to be recorded against a common reference axis. The amount that the individual's landmarks varies from the age standard is an estimate of the amount that he will vary from the mature-age standard after growth has ceased. For example, if a child's chin is I mm superior and 4 mm posterior to the age standard for those cephalometric landmarks, one can estimate that he will differ by the same amount from the location on the mature-face standard.

This method is used for each landmark until the final facial morphology is estimated. In a case analysis, it is not necessary to use a myriad of skeletal landmarks. The easiest number of landmarks is nine-Bolton point, Condylion, Gonion, Menton, B point, Pogonion, A point, maxillary incisor and maxillary first molar. However, others may be included or used to replace some of these.

Since the sex difference in the posteroanterior data was of no practical significance, the data on the males and females were combined. This template is used to assess the facial width measurements.

It would appear that the accuracy of the method could be increased if each skeletal landmark was individualized and compared to its mean and standard deviation. This may be accomplished by utilizing bivariate analyses in the form of age, sex, and growth-type specific confidence ellipses.

Each ellipse is constructed with its major and minor axes, which tend to be constant for each specific growth type past the age of 8 years. The axes of the ellipses are used as reference lines for measuring in the same manner as the mean occlusal line is utilized in the regular templates.

With the ellipse method, the landmarks are closer to the reference lines so that parallel and perpendicular measurements can be made more accurately. In this way, an individual is compared to his peers in terms of the mean position and the variability for a specific age, sex, and growth type. Each landmark has a different reference line so the individualization of landmarks is possible. The ellipse method is more accurate, but takes more time. In both methods, it is advantageous to use the chronological standard which most closely approximates the skeletal age of the extremely accelerated and severely delayed maturers.

Cranium - Mandible ---------------------Carnium - Maxilla - Mandible
Cranium - Maxila - Mandible
Fig 4 - The growth prediction methodology was based on co-ordinate measurements that were combined into these four areas with sella turcica as the origin.

 Fig 5 - The anterior maxillary horizontal component consisted of the five measurements indicated by the x's and their respective weights represented by the b's.

 Craniofacial growth prediction remains one of the ultimate goals in craniofacial biology and orthodontics. The records of 122 males and 111 females of the Serial Experimental sample were used to establish a method for predicting the mature face from earlier records.

A series of multiple regression equations were compiled to predict the horizontal and vertical changes of landmarks from coordinate data. This method used 38 independent variables from an early age for each of the 38 dependent variables of the mature face (Fig 3). These represent the horizontal and vertical co-ordinates of 24 cephalometric landmarks.

Fig 3 - The growth prediction methodology consisted of 38 horizontal or vertical co-ordinates of 24 points. These measurements were grouped into the 9 sets of measurements represented by the solid lines. As a result of the regression analyses, any one equation had fewer than 38 independent variables.

Percentage of variance of serial experimental and control samples predicted by Sex and Age specific serial experimental sample equations
Age 9 Predictors
B point
A point
Age 12 Predictors
B point
A point

The reference axis for the data model was sella-nasion, with sella as the axes origin. Sella turcica or the anterior cranial base is considered one of the most stable areas from which growth changes can be registered, and sella-nasion superimposition is practical and accurate for measuring increments. If another landmark had been used, the variance, covariances, and subsequent correlations would not have changed except for measurement error because the dimensions would have only been moved along the scale. Without a change in these factors, the regression results would have been the same.

The variables were horizontal or vertical components of anatomical landmarks. The geometric orthogonality of the horizontal and vertical components made it simple to investigate their effects, either together or separately. As well, the effects of the composite anatomical areas could be evaluated separately. The variables can be combined into the following four areas: superior vertical, anterior horizontal, inferior vertical, and posterior horizontal (Fig 4). These four groups divide the head into four distinct directional areas relative to the origin at sella.

The form of the multiple regression equations, is as follows: Y = a + blxl + b2x2... + bnxn; where Y is the predicted measurement (relative to sella), a is the intercept or axis correction factor, b are the individual regression coefficients or weights, and x are the individual measurements. Each weighted-average equation was the summation of the regression constant and the products of the length times regression coefficient for each horizontal or vertical measurement. The measurement products were combined into anatomical products, and subsequently combined as the products representing superior vertical, anterior horizontal, inferior vertical, and posterior horizontal values.

The biological accumulation (Fig 5) of the products was carried out in an indirect manner. One possible method was in the form of the summation of the regression coefficients that were involved in a particular measurement. As noted earlier, the form of the regression equation was: Y = a + b1x1 + b2x2... By summing the regression coefficients, the equation took the form of Y = a + b l (x 1- x 2) + (b2 + b1) x2 ... In this way, the model was a composite of many measurements, but very few of them originated at the origin of one of the axes.

Fig 5 - The anterior maxillary horizontal component consisted of the five measurements indicated by the x's and their respective weights represented by the b's.

The 122 male and 111 female cases of the Serial Experimental group were used to determine the accuracy of the predictions of the mature face. The residuals from the multiple regression equations were randomly distributed. The sex-specific percentages of variance that could be predicted for ages 9 - and 12-year-old Serial Experimental males and females ranged from 79 % to 94 % . (Table 3). The predictions of the Serial Control sample of 115 males and 83 females at the same ages showed prediction variances from 62 % to 87 %. The predictions were even better when some of the treated cases, delayed and accelerated maturers, and those without 20 year-old records, were deleted.



A total of 207 of the 312 children in the Serial Experimental sample received some form of orthodontic treatment. The type of treatment was divided into three categories: (a) Interceptive treatment included habit consultation, consultations concerning operative work, removal of supernumerary teeth, occlusal equilibration, slicing of mesial surfaces of deciduous cuspids, insertion of fixed and removable space maintainers, space regainers, labial shield, frenectomy, swallowing exercises, and extraction of deciduous and permanent teeth as part of serial extraction procedures, over-retention of deciduous teeth, and pathology; (b) interceptive plus compound treatment consisted of all forms of interceptive methods, and more complex treatment which included bite planes, expansion plates, monoblocs, labial or lingual arch wires, cervical headgear, and partial bands or full bands on upper and/or lower teeth; and (c) compound treatment included all the more complex forms of treatment without any interceptive procedures.

A thorough analysis of 108 male and 89 female cases of the Serial Experimental sample who received treatment, demonstrated that lhe majority of the children had received the correct type of treatment (Table 4). The remaining 10 treated cases dropped out of the study early and were not included. In retrospect, the ideal treatment plan, which included the mechanics and timing, was prescribed for 53 (49.1 %.) of the males and 52 (58.4 %) of the females. Of the 77 treated Class II cases, 35.7 % of the males and 37.1% of the females had ideal treatment mechanics and timing.


 Sex-specific retrospective treatment analysis of the 197 treated cases
Too Early
There were 44 males ( 40.7 %) and 27 (30.3 %) females who had received treatment too early. This sex difference resulted from insufficient consideration of maturation differences between, males and females at the time that the treatment was prescribed.

Of those who had treatment too early, activators represented 59.2%, cervical 16.9%, bite planes 12.7%, and the remaining 11.2% were made up of space maintainers and regainers, habit appliances, partial bands and inclined planes. In this group, there were 31 Class I and 40 Class II cases.

The appliance type was not related to the type of co-operation. However, with an increase in the treatment duration, there was an increase in the number of poor cooperators. Co-operation was a problem with 26.8 % of the males and 16.8 % of the females. Of the children who were treated too early, 38.6 % of the males and 14.8 % of the females demonstrated poor co-operation.

Generally the skeletal pattern of the untreated cases was better balanced than the extensively treated cases, before and after treatment. The Class II untreated cases appeared to have normal skeletal pattern at 6 years, and to have more normal growth at their last record. The malocclusion group that receved treatment had a less-balanced facial pattern before and after treatmen, even though they improved significantly during treatment, with major changes in the dento-alveolar area.

In the nontreated Class I cases, the upper and lower incisors were 2.0 mm. more protrusive, and B point and pogonion were 1.5 mm. more anterior than the Class I treated cases. This was possibly due to the extraction of premolars in the treated group. In the untreated Class II cases, the upper and lower incisors, B point and pogonion were 2.5 mm., A point 2 mm. and nasion 1 mm. anterior when compared to the treated Class II cases. When compared to the total sample of the Serial Experimental group, A point, B point, pogonion and upper and lower incisors were 1 mm. anterior in the total sample of the age 6 Serial Control group. This was possibly because the former contained more extraction cases.



The Burlington Growth Centre was established as a prospective clinical trial for orthodontic treatment and craniofacial growth. The data continue to be utilized for retrospective epidemiological and growth studies. In this way the Burlington Growth Centre is a practical data resource in both the fields of epidemiology and clinical trials.

This study was made possible by use of material from the Burlington Growth Centre, Faculty of Dentistry, University of Toronto, which was supported by funds provided by Grant 606-1349-40 under the Canadian National Health Research and Development Program.


1. MILLER, P.A; SAVARA, B.S.; and SINGH, J.J. Analysis of Errors in Cephalometric Measurements of Three dimensional distances on the Maxilla. Angle Orthodont, 36:169-178, 1966.

2. BAUMRIND, S.; MILLER, D.; and MOLTHEW, R. The Reliability of Head film Measurements, 3. Tracing Superimposition, Am . J. Orthodont, 70:617-644, 1976.

3. KROGMAN, W.M. and SASSOUNI, V. A Syllabus in Roentenographic Cephalometry, Philadelphia Center for Research in Child Growth, 1957.

4. SOKAL, R.R., and ROHL, F.J. Biometry, the Principles and Practice of Statistics in in Biological Research, San Francisco; W.H. Freeman and Co., 1969, pp. 526-532.

                                   ESTUDOS NO MATERIAL DO BURLINGTON - NOV 1999   

                                   Apresentação do material -   F. Popovich   

                                    Homenagem ao grande pesquisador