Development of bias control tool for observational health studies: A methodological study

Aim and hypothesis: The aim of this study is to develop an easy-to-use, comprehensive, cost effective and time efficient bias control tool for observational studies. Methods: This is a methodological study conducted between June 2018 – June 2020. Literature review to extract items, expert opinions, Delphi panel, construction of the framework and the content of the tool, and statistical analysis and reporting of the study were the main steps. "Expert Assessment Form" was used to evaluate expert opinions. Descriptive statistics were presented as mean (standard deviation) or median (minimum-maximum), frequency and percentage. After the the feedback of expert opinions; the items' with values of the Content Validity Ratio (CVR) that were equal or below the zero and other suggestions of experts related to the items were assessed in Delphi panels. Content validity study was based on consensus of experts. The name of the developed tool was decided to be Bias Risk Assessment Tool (BiRAT). It was considered to use the abbreviations BiRAT-CS for cross-sectional studies, BiRAT-CC for case-control studies and BiRAT-Co for cohort studies. Statistical analyzes were done with SPSS 26.0 (for MacOS) package program. Results: A total of 71 expert assessment forms were sent to 67 experts and 44 of them were received. 17 (38.6%) of the forms by e-mail, 14 (31.8%) of them by post (cargo), 11 (25.0%) of them by direct way (face-to-face), two (4.6%) of them by whatsapp application were sent to us. The numbers of the experts giving feedback for BiRAT-CS, BiRAT-CC and BiRAT-Co were 15, 18 and 11, respectively. 46.7% of them in BiRAT-CS were women and the median value of the total working time in the field of their expertise was 232.0 (111.0 – 255.9) months. 61.1% of the experts in BiRAT-CC were women and the median value of the total working time in the field of their expertise was 236.0 (137.0 – 354.0) months. 63.6% of the experts in BiRDA-Ko were women and the median value of the total working time in the field of their expertise was 232.0 (132.0 – 360.0) months. For cross-sectional study, a total of 77 items were created fort he assessment of the experts, and content validities of seven items (9.1%) were considered as insufficient. For case-control study, a total of 80 items were created for the assessment of the experts, and content validities of 14 items (17.5%) were considered as insufficient. For cohort studies, a total of 82 items were created fort he assessment of experts, and content validities of 12 items (14.5%) were considered as insufficient. As a result of the assessments made after the Delphi panels; 68-item BiRAT-CS, 69-item BiRAT-CC and 70-item BiRAT-Co were developed. Conclusion: BiRAT could be used to assess bias in observational studies (cross-sectional, case-control, and cohort). Since the usages of bias assessment tools were generally impractical, they should be used with mobile / online applications or artificial intelligence technologies for easier use and further development.