Clinical guidelines and the pursuit of reducing epistemic uncertainty. An ethnographic study of guideline development panels in three countries.
Wieringa S., Engebretsen E., Heggen K., Greenhalgh T.
PURPOSE: To explore, from a philosophy of knowledge perspective, the contribution of the guideline development process to reducing epistemic uncertainty in clinical decision-making - defined as the challenge of applying evidence to patients, dealing with conflicting information and determining the level of confidence in a medical conclusion. METHODS: Longitudinal ethnographic study of national guideline development panels. Fieldnotes were collected from 19 panel meetings in UK, Netherlands and Norway (~120 h of observation) between September 2016 and February 2019. Draft guidelines, review protocols and background material were collated (~200 documents). Data were analyzed thematically to gain familiarity and then theorized using concepts of knowledge development and use and clinical decision-making. RESULTS: Guideline development panels in all three countries wrestled with epistemic tensions - notably between the desire to "purify" an assumed external truth (for example by limiting included evidence to high-quality randomized controlled trials) and a more pragmatic and pluralist approach that drew on a wider range of evidence including qualitative research, real-world data, clinical experience and patient testimony. Detailed analysis of the process by which particular guideline recommendations were constructed allowed us to draw out the implications of these tensions for guideline users in clinical practice. CONCLUSION: Guideline development panels apply multiple - often conflicting - understandings of knowledge, inference and truth in an attempt to reduce epistemic uncertainty. Guidelines makers, clinicians, scientists and students should engage critically and reflexively with the philosophical assumptions that underpin guideline development and inductive inference to build capability to deal with clinical complexity.