Intention-to-treat analysis: Are we managing dropouts and missing data properly in research on orthodontic treatment? A systematic review

Introduction Intention-to-treat (ITT) analysis is an approach to managing dropouts and missing data in randomized controlled trials (RCTs). In this study, we systematically reviewed orthodontic RCTs to assess the frequency that an ITT analysis was carried out, to compare the number of trials that reported using ITT analyses with those that had truly used it, and to evaluate how dropouts and missing data were managed. Methods Systematic searches were conducted in electronic databases including Cochrane Oral Health's Trials Register (searched on November 30, 2016) and Cochrane Central Register of Controlled Trials (2016) in the Cochrane Library (searched on November 30, 2016), with no restrictions on language, publication year, or publication status. RCTs comparing orthodontic or orthopedic treatments, or comparing orthodontic or orthopedic treatment with a control group without intervention were included. A customized data collection form was created, piloted, and used to gather information from the selected studies. The data extraction was performed by 2 authors independently and in duplicate, with disagreements resolved by discussion with the third author. The studies were assessed for attrition bias. The data were grouped and classified according to 2 categorical variables: sample analyzed and missing data strategy. The results were reported in percentages and descriptively. Results From the 55 RCTs identified, 6 reported using an ITT approach. From these, only 1 study carried out a true ITT analysis (2%). From the 49 RCTs that did not report using an ITT analysis, 12 had used it (22%). The most used method of analyzing missing data was “completer sample” with 19 studies using this method (28%). “Full random sample” and “sufficient dose” were similar, with 13 and 11 studies, respectively (16% and 14%). The most frequently used missing data strategy for studies that did not conduct a true ITT analysis was “sample followed” with 30 studies (81%). For the studies that conducted a true ITT analysis, the most observed missing data strategy was “no dropouts” with 11 studies (79%). Conclusions Less than a third of the RCTs in orthodontics used an ITT analysis. There is a potential lack of understanding on dropouts and missing data management in research on orthodontic treatment.