<b>Background and objectives:</b> It has been argued that the binary framework frequently adopted to analyse test accuracy does not represent the clinical reality of diagnostic practice, and the recognition of an intermediate category of test result could make the utility of diagnostic tests more transparent. The objective of this thesis is to explore the value of moving away from the binary framework when evaluating and interpreting quantitative diagnostic tests. <b>Methods:</b> This thesis starts with an overview of the key arguments against dichotomising quantitative test results and a summary of some of the alternative methods proposed. Four distinct studies are then reported: 1) a systematic review of the methods currently used to evaluate the accuracy of quantitative cancer biomarkers, 2) a survey of GPs exploring preferences for threshold guidance 3) an evaluation of existing methods for identifying an intermediate range of test values, and 4) an assessment of the feasibility of applying these methods to the results of a meta-analysis. <b>Results:</b> The binary framework remains the most common method for evaluating the accuracy of quantitative tests, despite the survey of GPs indicating that a single threshold interpretation is less helpful than identifying rule-in and rule-out thresholds. Existing methods for identifying an intermediate range of values require some adaptation to incorporate the cost trade-offs relating to different outcomes but, given complete reporting at the primary research, could be applied to the results of a meta-analysis. <b>Conclusion:</b> The 2 x 2 diagnostic framework frequently fails to capture many of the realities and complexities of clinical research questions. Standardised methods that facilitate complete reporting of test accuracy in primary diagnostic accuracy studies are required to allow for greater flexibility when producing threshold recommendations further down the evidence pathway.

Au cours de ces dernières décennies, des progrès considérables ont été réalisés dans le domaine de l'oncogénétique, activité médicale basée sur les avancées scientifiques de la biologie moléculaire permettant de prédire le risque de survenue d'un cancer. Néanmoins, dans 80% des cas, les analyses génétiques initiales, réalisées auprès de patientes atteintes d'un cancer du sein et/ou de l'ovaire, ne permettent pas d'identifier une mutation délétère sur l'un des deux gènes BRCA. Dans ce cas, le résultat est dit « négatif », alors qu'il existe toujours un facteur génétique résiduel et le risque de développer un autre cancer justifiant la poursuite de la surveillance personnelle et familiale. En collaboration avec des médecins oncogénéticiens de la région Rhône-Alpes, nous avons interrogé les modalités de transmission de l'information dans ce contexte, en évaluant d'une part l'apport d'un document écrit remis aux patientes à l'issue de la consultation et d'autre part la nécessité du cadre de la consultation. Nous avons également tenté de mieux discerner la manière dont les informations orales sont perçues par les patientes. Pour ce faire, nous avons interviewé 88 patientes au cours d'un entretien clinique de recherche et 172 patientes ont répondu à un questionnaire. Les résultats obtenus ont montré : - l'utilité de délivrer un document écrit résumant les informations orales transmises par le médecin ; - le cadre indispensable de la consultation pour délivrer le résultat négatif ; - l'existence d'un message paradoxal, dans la mesure où le résultat « non concluant » est annoncé comme négatif, rendant la compréhension de l'information plus difficile. / Over the last decades, major progresses have been realized in oncogenetics. This medical activity based on scientific advances in molecular genetics predicts cancer risk. However, no deleterious mutation is detected by BRCA1/2 genetic testing for the majority (80 %) of women who suffer from either breast or ovarian cancer. Under this condition, the result is considered as “inconclusive” although there remains an increased risk for developing breast and/or ovarian cancer. The failure to identify a mutation in a diagnostic test does not rule out the existence of a predisposing factor. Surveillance should be pursued for both the patients and their families. In association with genetic counselors of the region Rhône-Alpes, we considered useful to address the question of information delivery in the context of an inconclusive result. We investigated on one hand the benefit from a written document delivered to patients upon the appointment, and on the other hand the usefulness of a disclosure session. We also investigated how patients experienced the oral information given during the counsel session. Therefore, we conducted semi-structured interviews with 88 women while 172 women answered a questionnaire. The observations demonstrate : - a written document, summarizing the oral information delivered by the oncologist, is useful for patients with an inconclusive result; - a disclosure session is absolutely necessary to announce the inconclusive result; - the presence of a “paradox” when the inconclusive result is delivered as a negative one, increasing the degree of comprehension of information more difficult.