86 chapter 2 Quality assessment Quality and applicability of the included articles were appraised using the QUADAS-2 criteria. Discordance on various QUADAS items was resolved through consensus between the two investigators. The consensus QUADAS assessments are presented (Figure 2). Most risks of bias concerned the flow and timing: patients who were lost during study follow-up, and most importantly high numbers of missing reference standards, which varied greatly among the different tests. Most unclear risks of bias existed in the patient selection domain, as many studies provided only a summary description of their inclusion procedure. There were little concerns regarding the applicability of studies. Figure 2. Consensus QUADAS-2 assessments Risk of Bias Applicability Concerns patient selection index test reference standard flow and timing patient selection index test reference standard 0% 100% 50% percentage of studies QUADAS-2 low high unclear Publication bias The included studies for all index tests were assessed for publication bias. Varying degrees of visual funnel plot asymmetry were observed (Figure 3), but corresponding results of the Egger’s regression-based tests were not significant. Presence of publication bias was not suggested for any of the index tests.
RkJQdWJsaXNoZXIy MTk4NDMw