We thank Drs Perinetti and Contardo for reading our article carefully and the effort they made to apply another analysis on our data. However, we would like to alert our colleagues to various mistakes in their analysis and interpretation. We cannot agree that “dental maturation is not a reliable indicator of the pubertal growth spurt.” If we look at the frequency table they built from our data, we can clearly see that the conclusion of our paper is correct: “An association exists between the dental mineralization stages and the skeletal maturation periods in the studied population.”
The great advantage of the analysis we used is that for every stage of maturity, the analysis takes into account the maturation stages of other teeth. This is important because patients have maturity variations of the different teeth. Therefore, it is extremely important to consider each person’s teeth set in the statistical analysis. Another important aspect is that our analysis considers that the response variable is an ordinal variable. These are advantages of ordinal multinomial logistic regression. The dichotomization of the dependent variable compromises the data analysis because there is a loss of information by agglutination of categories with disregard for the ordination between them.
These colleagues based their analysis on an article whose response variable was dichotomous (with and without obstructive airway disease) and assessed only one independent variable (smoking habit). Since in our case the response variable is ordinal (3 levels of maturation), we cannot calculate rates by comparing the probability of each growth phase with the combined probability of other phases as presented by our colleagues. In fact, there is an error in their methodology and interpretation. When the likelihood of each growth phase is compared with the likelihood of other combined phases, the result is thinned. Obviously, in the intermediate group (peak), as they performed in their analysis, the likelihood ratios are lower than if they were calculated for each of the 2 other stages of growth. In addition, we must be cautious when using an arbitrary threshold so high. In this case, a threshold greater than 10 was established by Jaeschke et al as strong evidence in diagnostic tests in most circumstances, but we cannot ignore a likelihood ratio of 5, for example.
Finally, we are sure that our results, and even the mistaken results of our colleagues, support our conclusion: that an association exists between the dental mineralization stages and the skeletal maturation periods. The tooth analysis can contribute to the request for hand-wrist radiography at an ideal time. Therefore, the evaluation of the dental mineralization stages can promote reductions in the requirement of hand-wrist radiographs, because dental practitioners can avoid the prepubertal phase (when they find teeth in stages D or E for girls and stages E or F for boys, especially the second molars) and the postpubertal phase (when they observe teeth in stage H for both sexes). However, for a patient with teeth in the intermediate stages and probably in the peak of the growth period, the professional should request hand-wrist radiographs to indicate the patient’s specific growth phase.