Material and methods

We used radiographic images of 124 children (8-12 years old) with up to 4 missing teeth (55 boys, 69 girls) that we compared with the images of 676 reference children (365 boys, 311 girls) from the Nijmegen Growth Study and the Generation R Study in Rotterdam in The Netherlands ( Table I ).

The Nijmegen Growth Study is a mixed longitudinal population-based cohort study conducted from 1971 to 1976 at the Radboud University Medical Center in Nijmegen, The Netherlands. This study included 3 cohorts: children were enrolled at 4, 7, and 9 years of age and followed until they were 9, 12, and 14 years of age. We used only 1 radiographic image per child taken between 8 and 12 years. If a child had more than 1 radiographic image available, we selected the one taken at the age that was closer to the mean age of the Generation R Study sample. In total, 7 children with hypodontia and 203 reference children were included from the Nijmegen Growth Study, with an average age of 9.39 ± 0.32 years. Before the inclusion of the participating children in the Nijmegen Growth Study, signed consents were obtained from the parents.

The Generation R Study is a population-based cohort study from fetal life to adulthood, established in Rotterdam, The Netherlands, at the Erasmus University Medical Centre. From the fourth data collection phase, we used data from 117 children with hypodontia and 473 healthy children with a mean age of 9.67 ± 0.40 years. The study was approved by the medical ethics committee of the Erasmus Medical Centre (MEC-2012-165) in Rotterdam. At the start of each phase, mothers and their partners received written and oral information about the study and were asked for their written informed consent.

Hypodontia in children was assessed from their dental panoramic radiographs. One dentist (B.D.) determined the number and position of the missing teeth for each subject. In total, 124 children had hypodontia ( Table I ). Of those, 39 children had anterior hypodontia (tooth agenesis of incisors and canines), and 85 children had posterior hypodontia (tooth agenesis of premolars and molars). Also, we classified hypodontia based on the jaw in which the tooth was missing. Our sample consisted of 29 subjects with maxillary hypodontia, 86 subjects with mandibular hypodontia, and 9 subjects who had missing teeth in both jaws. No child had more than 4 missing teeth or a combination of anterior and posterior hypodontia.

We selected 14 landmarks on cephalograms ( Table II ); from these cephalometric points, we measured 12 angular parameters, 1 distance, and 2 derived proportions. The mean values of the cephalometric parameters for boys and girls are given in Tables III and IV , respectively. Lines Mx (palatal plane) and Mn (mandibular plane) were obtained by connecting points ANS and PNS, and Go and Me, respectively. Line Ui passes through the axis of the maxillary central incisors by connecting points Is and Rs, and line Li goes through the axis of the mandibular central incisors by connecting points Ii and Ri. Before each measurement, the image was recalibrated depending on the magnification of the cephalometric radiograph. We included the most frequently used cephalometric parameters from previous studies that investigated cephalometric differences between hypodontia and reference groups. In both the Nijmegen Growth Study and the Generation R Study, cephalometric landmarks were digitized from which linear and angular cephalometric parameters were calculated. Cephalometric points in the Generation R Study were digitized by a dentist (S.V.) using Viewbox software (version 4.0; dHAL Software, Kifissia, Greece).

We applied a PC analysis to combine correlated cephalometric parameters into a new set of uncorrelated PCs, each representing a distinct craniofacial pattern. A detailed description of the PC procedure used in this study is provided in the Supplemental material . Briefly, we explored first the intercorrelation among cephalometric parameters using Pearson correlation ( Table S1 ). Secondly, a standardized score was created for each PC with a mean value of 0 and a standard deviation of 1. If a cephalometric parameter had a loading greater than 0.5 for a component, we interpreted the increase or decrease of that component as an increase or a decrease of that cephalometric parameter. If a cephalometric parameter had a loading lower than −0.5, an inverse principle was applied. Cephalometric parameters that are included in 1 PC are bolded in Table S2 .

Interobserver agreement for the cephalometric measurements in the Generation R sample was tested using intraclass correlation on a subsample of 20 randomly selected cephalograms scored by 2 independent observers (S.V. and B.D.). Values of the correlation coefficient ranged from 0.699 to 0.972. The reliability of the measurements done in the Nijmegen Growth Study was published earlier, and it was considered highly reliable. We added study sample as a covariate in the linear regression analysis to adjust for the potential measurement differences between the Nijmegen Growth Study and the Generation R Study.