Dunja Dreesens

96 Conclusion The development of guideline recommendations is an interactive human process that requires a range of knowledge andexperience including, but not exclusively, knowledge from frequency-based research, such as clinical trials. As in the clinical encounter, appraising and including different types of knowledge in guideline development should be used to make better inferences to guide decisions, but in practice, arguments are used to exclude some kinds of knowledge for a range of reasons, including concerns about introducing bias in frequentist reasoning. In this paper, we present important epistemological reasons to appraise and include a (wide) variety of different types of knowledge to highlight important aspects of guideline development that await further exploration and practical suggestions. We acknowledge that appraising and including knowledge from a different variety of sources is likely to be complex and ongoing. Discussions about purpose, reasoning and integration in guideline development will continue. A simple set of tools or methodological quick fixes are unlikely to suffice, and developing criteria for appraising and integrating different knowledge will remain a challenge. However, we believe that much can be done to help guideline developers improve this – now often implicit (8) practice that is central to their work. Capacity-building workshops that confront implicit forms of reasoning are one example. AID Knowledge runs such workshops annually at GIN conferences. They help to strengthen ties between guideline developers who are concerned about the increasingly rigid methodological constraints on guideline methods at the expense of fostering epistemic sensibilities. It is important for guideline developers to feel they are part of a community of practice that encourages epistemic skill development, rather than a hierarchical community where superior guideline methods are defined by a small group of experts. This will help to keep guideline development innovative and diverse. Acknowledging that dominant frequentist methods are excellent for some questions but do not fit all knowledge needs is the first step to implementing different kind of reasoning in guide- line development. How to address the diversity in methods for different kinds of questions should be among the top guideline research priorities. Glossary Inference To reach a conclusion from premises (points, reasons, evidence, etc.) Induction A kind of inference that is risky. Even when the premises are correct, the conclusion may not be. This in contrast to deductive reasoning where the conclusion is necessarily correct. The problem of induction A concern in philosophy whether inductive inference is justifiable. Evasion A certain way of reasoning to deal with the problem of induction. Chapter 5