Mariska Tuut

Applying GRADE for diagnosis 29 2 Introduction Clinicians use tests to ascertain or reject a clinical diagnosis [1]. The clinical value of a test depends on various elements: the patient population characteristics (e.g. prevalence of the disease), test characteristics (e.g. sensitivity and specificity) and its downstream consequences on patient-relevant outcomes (e.g. test burden, natural course of the disease and management following the test results) [2]. Since direct evidence evaluating the impact of tests on patient important outcomes (diagnostic randomised trial) is scarce, different types of evidence (e.g. for diagnostic accuracy and management effectiveness) need to be assessed and linked. Clinicians often have a limited ability to assess the value of a test in clinical practice [3, 4]. Therefore, clinical practice guidelines (CPG) have been developed to provide decision support to clinicians and patients[5]. The GRADE approach for diagnostic tests and test strategies facilitates this process by linking the elements of a testtreatment strategy and assessment of the certainty of the evidence for each element [6-8]. It is challenging to appropriately evaluate diagnostic tests (e.g. assessing the certainty of the evidence, including patient-important outcomes in evaluating test accuracy) [9, 10]. In this study, we aimed to identify the challenges of applying GRADE for diagnosis for all elements of the test-treatment strategy. We assessed the certainty in the evidence in an illustrative example and proposed solutions to overcome the barriers. This study may serve as an example for systematic reviewers and guideline developers. Methods Clinical question The illustrative example is the clinical question: what is the value of specific immunoglobulin E (sIgE) blood testing as an add-on test to history taking (I) compared to history taking alone (C) in patients suspected of having allergic rhinitis (AR) in primary care (P), with relief of nasal or ocular symptoms as critical outcomes (O) [8, 11]? Concentration, sleep problems, work/school absence and quality of life (QoL) were considered important outcomes [12]. Consequences of true positive, true negative, false positive, false negative, and failed test results were discussed. We formulated PICOs for each element of the test-treatment strategy (see table 1). Search strategy Detailed methods for searching and assessing evidence for each evidence element are presented in table 2. We searched Medline and Embase databases to retrieve relevant evidence (Appendix 1). We searched for publications from 1998 to 11 January 2019

RkJQdWJsaXNoZXIy MTk4NDMw