A new approach for explicit judgments of values and preferences: define the benefit required for an intervention

Tags: Oral
Helsingen LM1, Siemieniuk R2, Vandvik PO3, Zeng L2, Bretthauer M3, Agoritsas T4, Guyatt G2
1Oslo University Hospital, 2McMaster University, 3University of Oslo, 4University Hospitals of Geneva

Background: judgments about values and preferences, implicit or explicit, inform all clinical practice guidelines. When, as is frequently the case in screening for cancer, a single critical outcome exists, a potentially useful framing of the key question is: 'Given the expected harms and burdens, how much benefit do people in the target population require to choose an intervention?' Although this question is crucial, it is typically not addressed explicitly.

Objectives: we describe implementation of a seldom-used method for making explicit judgments of values and preferences. As a worked example we use the approach taken in a recent guideline on colorectal cancer screening (BMJ Rapid Recommendations).

Methods: before examining current best evidence, the guideline panel considered harms, burdens and practical issues associated with four screening options versus no screening: Faecal immunochemical testing (FIT) every year or every two years, sigmoidoscopy and colonoscopy. Considering these undesirable consequences, the panel completed surveys assessing the benefits at 15 years that typical people would require to undergo screening. The survey results informed the definition of thresholds of benefit below which most people would decline screening, and above which they would choose screening. Bearing these thresholds in mind, the panel examined the evidence and issued their recommendations.

Results: the panel inferred that at least half of people in the target population would require a colorectal cancer mortality reduction of 5 per 1000 over a 15-year period to undertake FIT-screening, and a reduction of 10 per 1000 for sigmoidoscopy or colonoscopy. They clarified that a difference in colorectal cancer mortality reduction of 10 or more per 1000 would prompt them to recommend colonoscopy or sigmoidoscopy over FIT, and a difference of 5 or more would prompt recommending FIT every year over FIT every two years and colonoscopy over sigmoidoscopy.

Conclusions: setting a threshold for the required benefit before reviewing the evidence can minimize the influence of panelists' pre-conceived beliefs regarding appropriate recommendations and enables more explicit and transparent judgments regarding values and preferences. The approach is suitable for decisions in which one key beneficial outcome is weighted against potential harms and burdens. Cancer screening represents good examples in which disease-specific mortality is the most important expected benefit. This approach could facilitate efficient and coherent recommendation development across a range of individual prognosis.

Healthcare consumer involvement: three people with colorectal cancer screening experience took part in the guideline development process, including defining thresholds of the required benefit.