The prevalence of and factors influencing statistical testing of baseline characteristics in randomized controlled trials published in high-impact orthodontic journals.

INTRODUCTION:When proper randomization has been undertaken, statistical testing of baseline characteristics between participants in trial arms in randomized controlled trials (RCTs) is not required. This investigation aimed to assess the prevalence of statistical testing of baseline differences in orthodontic RCTs. Factors influencing the undertaking of this analysis were explored. METHODS:Orthodontic RCTs published between January 1, 2017 and December 31, 2021 in 5 orthodontic journals were identified. To determine if statistical testing of baseline differences had been undertaken, each article was reviewed in detail to identify the reporting of P values and the term "significant difference" in the table of characteristics, the table legends, and the results section of each included RCT. Trial characteristics at the RCT level were extracted. Frequency distributions were calculated for the included trial characteristics. Significant predictors from the univariate analysis were used to construct a multivariable Bayesian logistic regression model. RESULTS:One hundred and thirty-two RCTs were analyzed. Significance testing of baseline characteristics was undertaken in 50% (66/132) of RCTs. At a journal level, significance testing at baseline was infrequently undertaken in RCTs published in the American Journal of Orthodontics and Dentofacial Orthopedics (AJODO). Compared with 2017, RCTs published in 2018 (odds ratio [OR], 5.57; 95% credible interval [CrI], 1.33-25.69), 2019 (OR, 17.82; 95% CrI, 4.41-82.11), 2020 (OR, 6.48; 95% CrI, 1.72-27.12), and 2021 (OR, 3.24, 95% CrI, 0.81-14.01) had higher odds of significance testing at baseline. RCTs published in the European Journal of Orthodontics (OR, 5.31; 95% CrI, 1.79-17.04), Progress in Orthodontics (OR, 5.00; 95% CrI, 0.97-28.43), Orthodontics and Craniofacial Research (OR, 6.49; 95% CrI, 1.04-46.10), and Angle Orthodontist (OR, 12.30; 95% CrI, 3.27-51.44) had higher odds of significant testing at baseline testing compared with AJODO. CONCLUSIONS:Statistical testing of baseline differences is common in orthodontic RCTs. Trials published in AJODO had the lowest incidence of statistical testing of baseline differences. RCTs published between 2018-2021 had higher odds of significance testing at baseline than in 2017. Per the consolidated standards of reporting trials guidelines, this practice should be discouraged as it can be misleading and unnecessary.