The article by Barros in the November issue is extremely important (Barros SE, Chiqueto K, Janson G, Ferreira E. Factors influencing molar relationship behavior in the mixed dentition. Am J Orthod Dentofacial Orthop 2015;148:782-92). It provided important additional evidence about molar relationships after the establishment of the early permanent dentition. This letter is to praise this study because of its importance to dentists who treat children during the mixed dentition transition. However, I think it is important to express concerns regarding the use of combined samples, which might be problematic.
In this study, a total combined sample size of 80 children (44 boys, 36 girls) was selected from a pool of 1212 untreated children from the Michigan Growth Study and the Bauru Dental School Growth Study in Brazil. It is unclear whether those 80 subjects were representative of the original samples and the underlying populations. The “Material and methods” section did not include a description of the populations of these 2 growth studies or references to detailed descriptions. Furthermore, no explanation was provided for the methods used to adjust for sample characteristics in the statistical analysis.
An important consideration when combining samples is the variance. The weighted pooled sample variance provides a higher precision estimate of variance and greater power than the individual sample variances. If the variances of the 2 samples are not the same (ie, if they are statistically significantly different), then a weighted average of the variances should be used. It is not clear how this issue was considered in the analysis of the study of Barros et al. From my extensive review of the literature, I found several studies that used combined samples. However, from my perspective, the heterogeneity was not as critical as in the study of Barros et al, since the authors of the other referenced articles combined homogenous populations. For example, in the studies of Bishara et al, all study subjects were white Americans who were patients of the University of Iowa Hospital and Clinics. It is well known that the cephalometric norms of white Americans, for example, are different from those of African Americans. Thus, combining 2 different ethnic samples in orthodontic studies that include cephalometric measurements should be done with caution, after consideration of race, ethnicity, and statistical issues, as outlined above. I think it would be important to the readers to know what adjustments were made, if any, concerning the samples. If none were made, then I think that this is a limitation of the study, since it might affect the validity of the statistical analysis.