An intercenter comparison allows access to adequate samples for investigating clinical outcomes and international variations in treatment outcomes and growth adaptation . However, intercenter studies cannot eliminate susceptibility or proficiency bias as the patients are drawn from different populations and the surgeons are inevitably different, but the patients from the three centers in the present study were treated by a limited number of surgeons according to a strictly defined and consistent protocol (see Table 2). Nevertheless, intercenter studies are not easy to perform. The variability in record taking and treatment protocol, even within the same center, as well as many co-factors such as clinician skill, proficiency, and the possibility of adapting a treatment procedure to the expected prognosis, make intercenter studies difficult. Even if the research evidence for retrospective longitudinal studies is considered to be rather weak, it has the advantage of recruiting consecutive cases for consistent evaluation [10, 11]. For the present study, we were able to include 148 patients with CBCLP who were followed longitudinally over a 6-year period, which is the largest sample reported in the literature to date. Only a few studies with very small samples cover the same age period longitudinally (Table 1).
In order to reach a consensus on data collection for further research purposes, the Eurocleft project has specified the ages for recording cleft lip and palate patients. Cephalometric radiographs were recommended at the age of 10 years . In the present study, the CBCLP patients were born before 1996 in order to have radiographs at the two target ages and is the reason why our age groups were not in accordance with the Eurocleft recommendations published later .
Three-dimensional cephalometry is the latest tool, but 2D cephalometric analysis is still the classic tool for describing facial growth and development in patients with cleft lip and palate. Because of concerns about the radiation dose of multi-slice or cone-beam computer tomography, it will probably continue to be the evaluation tool for longitudinal studies on facial growth and treatment outcome. However, in addition to the fact that 2D cephalometry is a two-dimensional representation of three-dimensional structures, cephalometric measurements have an inherent method error that varies depending on the radiographic projection, measuring system, type of landmark, and observer. Differences in the magnitude of the measurement error are caused by the precision of landmark definition and the amount of noise from adjacent structures. In young patients with cleft lip and palate, the identification of cephalometric landmarks is even more difficult due to abnormal anatomy, especially for the localization of the landmarks point A, anterior nasal spine, and posterior nasal spine . As described by Hotz and Gnoinski , point A is difficult to locate in young individuals because of the tooth germs molding the anterior contour of the maxilla. The most difficult age for examining radiographs in patients with a cleft is the period before shedding of the incisors, as all of the above-mentioned problems occur in this period of time. In our study, the intra-observer measurement error showed a systematic difference for one of 20 variables in the 6-year group and one in the 12-year-old group (see Table 4).
At 6 years of age, all patients (at all centers) with CBCLP showed a large SNA angle with retroclined upper incisors. This finding should not be interpreted as a prognathism of the entire maxilla, but the large SNA angle is the result of a forward positioning of the premaxilla in bilateral cleft lip and palate. Cephalometric findings at an even earlier age than examined in our study have shown an extremely protruding premaxilla with a short maxilla of reduced posterior height, a short mandible, and bimaxillary retrognathia with a more vertical facial growth pattern . The protrusion of the premaxilla in the 6-year olds from all three centers was similar to the recently published results of a well-documented longitudinal study (from age 5 until the end of growth) on the treatment outcome of Zurich’s treatment protocol in 5-year olds .
In the following 6-year period, the protrusion of the premaxilla diminished similarly for all three groups but occurred most in the Nijmegen group (see Table 6 and Fig. 2a, b). This pattern was also seen for the ANB angle and the corresponding soft tissue variable Snss. This change probably reflects the change brought about by the osteotomy of the premaxilla, which was performed in all patients at Nijmegen with the bone grafting procedure at a mean age of 9 years and 9 months. The direct effect of this operation is a better sagittal position of the premaxilla and an improved inclination and vertical position of the upper incisors, as well as reconstruction of the alveolar process to a normal height and width to create optimal conditions for canine eruption [17, 18]. However, whether the premaxillary surgery will result in impaired forward growth of the maxilla in the long run remains to be seen. In a preliminary cephalometric study that included seven patients from the present study, patients were followed longitudinally from 6 to 20 years of age for their final facial growth . At the age of 20, osteotomy of the premaxilla at a mean age of 13 years and 3 months was not found to have been detrimental to facial growth. Comparable results were found by Padwa et al.  and Geraedts et al. , who showed that a protrusive premaxilla can be positioned after the age of 6 to 8 years without deleterious effects on midfacial growth. However, the final outcome for the present sample remains to be investigated when growth has ceased.
In the Nijmegen patients, NS-NL decreased and NL-ML increased, indicating a counter clockwise rotation of the premaxilla. This pattern differs from that of the other centers and can probably be attributed to the surgical repositioning of the premaxilla in the CBCLP patients at Nijmegen before the age of 12.
The regression analysis (Table 7) showed that most of the hard tissue variables and all soft tissue variables at 12 years are explained by the relevant cephalometric values at 6 years of age. The R square numbers show that you can explain approximately 50% of the variability in 12-year values. The only variables that are not predictive are the ones related to the maxillary incisors, which could be expected as patients differ with respect to their dental developmental stage when the cephalograms were made. Gender did not play a significant role in explaining the cephalometric outcome at 12 years of age.
Many components that are difficult to identify are involved in the final outcome of cleft lip and palate patients. In addition to the growth variability between individuals and racial groups, drawing the line for the ideal treatment protocol is difficult as the treatment protocols of the three centers have primary differences in the early management of clefts, infant orthopedics, the type of lip repair (one-stage or two-stage approach), early versus late hard palate closure, and premaxillary osteotomy at the age of 9 years (Table 2). The developmental heterogeneity of individuals between centers is also an important factor. In a comparative study of cephalometric values among five centers, Nijmegen had significantly more Class II skeletal patients compared with all other centers . In the present study, we also noticed that the Dutch children had a significantly more retrusive mandibular growth pattern than the Scandinavian children.