164 Chapter 7 Assessment of methodological quality We will use the QUADAS‐2 (Quality Assessment of Diagnostic Accuracy Studies) tool for the assessment of methodological quality of all included studies (35). We will evaluate the four domains in QUADAS‐2 (patient selection, index test, reference standard, and flow and timing) for risk of bias, and the first three domains for concerns regarding applicability. For comparative diagnostic test accuracy studies, we will use the QUADAS‐C tool (36) (Supplementary Table 5). Two review authors (RT and KR) will evaluate the methodological quality of all included studies independently; we will resolve disagreements by involving a third author (CK or RW). We will pilot test both tools and only repeat this if it demonstrates problems during the first round. Statistical analysis and data synthesis Diagnostic test accuracy for each index test We will perform the analysis for each index test separately. We will perform random‐ effects meta‐analysis in a bivariate model (37). For each index test, we will present pairs of sensitivity and specificity with their 95% confidence intervals (CIs) for each study, as well as the pooled sensitivity and specificity, in a forest plot. We will then present the summary receiver operating characteristic (SROC) plot with summary points, and incorporate the seven domains of the QUADAS‐2 tool into this SROC plot. Unit of analysis We will analyse the data on a participant level in the analysis, as this is more clinically relevant. However, if per participant data are not available and per tube is the unit of analysis in the majority of included studies, we will consider performing analysis on a per tube basis. Comparative diagnostic test accuracy for different index tests Direct comparison We will include studies directly reporting two or more index tests compared with the reference standard in the primary analysis of comparative diagnostic test accuracy of different index tests. We will add a covariate for type of test in the bivariate model to compare the differences in test accuracy, and perform a likelihood ratio test to compare models (38). We will also present linked SROC plots, linking estimates of two different index tests from the same studies.
RkJQdWJsaXNoZXIy MTk4NDMw