Gender-specific treatment effects and outcomes reported in orthodontic research. A cross-sectional empirical study

Abstract

Aim To identify practices of assessment of gender effects in research articles in orthodontics and detect whether there were significant differences in the treatment effects on outcomes according to gender. Materials and Methods Four major orthodontic journals were sought over a 3-year period to identify publications which included assessment of gender effects on outcomes in their reporting. Data were extracted on the following characteristics: journal, year of publication, region of authorship, and study design. For the studies including reporting of gender effects, whether a significant effect existed was further documented. Additionally, for these studies, data were extracted on population, sample size per gender, treatment, comparison, outcome type, and nature and whether gender analysis was based on subgroup testing or included as a main effect. Descriptive statistics, cross-tabulations, univariable, and multivariable regression models were utilized as appropriate. Results A total of 718 research articles were eligible for inclusion out of a pool of 1,132 screened articles. Of those, 95 reported on any type of analysis on gender effects (95/718; 13.2%). In the 95 studies that reported assessment of gender effects, it was clear that the majority did not detect significant gender-related differences across the documented outcomes (range of frequency distribution for significant gender differences across all outcomes: 0–50%). Twenty-two articles overall (22/95; 23.2%) described a significant gender effect classified by outcome, 12 favoring female and 10 favoring male participants. Patterns of efficacy and adverse outcomes were schemed either favoring female (root resorption: 4/10; 40.0%, periodontal outcomes: 3/11; 27.3%) or male (cephalometric/growth changes following orthodontic treatment: 4/17; 23.5%) patients across the 22 studies with significant effects. Appropriately designed and adequately powered statistical analyses, with gender effect assessment as a main effect in a multivariable regression model was associated with 6.53 times higher odds for identifying significant gender effects (OR = 6.53; 95% CI: 2.15, 19.8; P = .001). Conclusions A very small proportion of research studies included gender effect assessment in their analyses. Of those, a quarter described significant effects. Nevertheless, careful analysis planning and strategies should be prioritized to allow for any meaningful interpretation.