40 Chapter 2 Discussion Summary of the main results By applying GRADE for diagnosis, we systematically evaluated elements of a diagnostic test (i.e. diagnostic accuracy, test burden, management effectiveness, natural course and the link between test result and clinical management). During this process, we faced challenging issues and suggested solutions to resolve them. This study can therefore serve as an illustrative example for evaluating a diagnostic test in the context of CPG development, considering effectiveness as well as efficiency. For the sIgE-test, the results suggest that it is very uncertain whether it contributes to quality of life and to reducing AR symptoms and work/school absence. The diagnostic accuracy was quite high, but with very low certainty. And the downstream consequences were very uncertain as well. A CPG panel probably would not recommend the routine use of sIgE blood testing for the diagnosis of AR in primary care. The main challenge in assessing the overall certainty of evidence was the lack of highquality evidence for the various elements of the test-treatment strategy. For most outcome measures of the elements in the evidence chain, we found very low-quality evidence. Another challenge was the time needed to systematically evaluate the complete pathway. Consulting CPG panel members, including patient representatives, can help save time by selecting elements of the test-treatment strategy for which a systematic review of the evidence must be carried out and others for which one can rely on other CPGs or expert opinion. If this selection process is motivated and described explicitly, the certainty of the guideline will not be affected [46]. During CPG development, a guideline panel should determine which outcome measures should be included. Panel members can also advise on the methodological approach per element of the test-treatment strategy and about the necessity of performing systematic reviews. These discussions in the CPG panel are essential to good guideline development. Gopalakrishna and colleagues applied the GRADE approach to three Cochrane reviews to evaluate the applicability of the GRADE approach for diagnosis [9]. They found challenges in formulating the question and applying the GRADE criteria. For example, assessors in this study experienced difficulties in judging the risk of bias in relation to the QUADAS criteria. The authors also identified issues with indirectness, inconsistency, imprecision and publication bias, all related to diagnostic accuracy.
RkJQdWJsaXNoZXIy MTk4NDMw