44 | Chapter 2 To avoid missing any relevant record, we manually screened 50% of all records, focusing on keywords ‘regret,’ ‘conflict,’ or ‘shared-decision making’ in title or abstract, before letting Rayyan rate the remaining records. Next, we manually screened 2000 records in descending rating order, after which ratings were recalculated for the remainder. Of those, we manually screened all 2.5 star records, and a random sample of 10% of the lower-rated records. All records were screened by two mutually blinded researchers (AH and DJ or CA). Included articles were labeled by topic (DR or DC), study design and tumor location. Discrepancies were solved by consensus. Data extraction Data extraction included author, year of publication, study type, study population, tumor location(s), tumor stage, treatment modalities, methods of measurement for DR and DC, and all reported data regarding DR and DC. Since studies used different scales and measures for point estimates and dispersion (due to different underlying distributions), which would prohibit useful meta-analysis, we used a dichotomous outcome (present yes/no) for the meta-analyses. Where needed, authors were contacted for data regarding the prevalence of DC and/or DR in their sample. Quality and Risk of Bias Appraisal All appraisals were done by two mutually blinded researchers (AH and DJ or CA) and were discussed until consensus. For quantitative research designs, we used the CLARITY checklist for assessing Risk of Bias (RoB) in Cross-Sectional Surveys23. This 5-item tool addresses population, response rate, missing data, clinical sensibility, reliability, and validity of the survey instrument. We rated the RoB for the reliability and validity items as low if a study used a scale that had been (previously) validated. The CASP Qualitative Studies Checklist was used for qualitative studies24. This 10-item checklist consists of three sections pertaining to the validity, results, and the extent to which the results are valuable for the context in which the study is used. Data synthesis and statistical analysis Data from qualitative studies was analyzed through inductive content analysis25. Themes as reported were extracted and coded by AH. In a group meeting (AH, CA, DJ), similar but differently worded themes were combined under a single term. Overlap between themes were examined, and themes describing different, but related concepts were merged into a higher-order overarching theme. Finally, we examined overall saturation of the data.
RkJQdWJsaXNoZXIy MTk4NDMw