Journal of Orthodontics | 2019
Confounders in epidemiological associations
Abstract
The following description of a retrospective non-randomized study is given within a publication in a scientific journal: “A total of 54 consecutive patients (24 male / 28 female; mean age: 14.1 years) treated with sequential thermoplastic aligners during the last 12 months were identified from the archive of a private practice. They were compared to 52 consecutive patients treated with conventional fixed appliances during the same period, who were matched for age, sex, and case severity with the patients in the aligner group. The total duration of treatment in months was extracted from the patient files by a third party not involved in any way with their treatment. Initially, descriptive statistics were calculated for all patient characteristics and for the study’s primary outcome (treatment duration), consisting of means and Standard Deviations (SDs). Student’s t-tests for independent samples and chi-square tests were performed on patient age, sex, and case severity to confirm that the two groups were matched. Finally, a t-test for independent samples was performed on treatment duration to assess any differences between the aligner and the fixed appliance group at the 5% level. The authors of the study give the following table (Table 1) in their Results section and conclude that (i) the two groups were adequately matched, since no statistically significant difference was found for any baseline difference and (ii) aligners and braces are equally efficient, since no statistically significant difference was found for treatment duration.