Evidence, values and context preferences to help mitigate disputes and enhance the applicability of guideline recommendations to practice

Article type
Authors
Parmelli E1, Capobussi M2, Gonzalez-Lorenzo M2, Puljak L3, Riva N4, Squizzato A5, Moja L6
1Department of Epidemiology Lazio Region-ASL Roma1
2Department of Biomedical Sciences for Health, University of Milan
3Cochrane Croatia, University of Split School of Medicine
4Department of Pathology, Faculty of Medicine and Surgery, University of Malta
5Research Center on Thromboembolic Disorders and Antithrombotic Therapies, Department of Medicine and Surgery, University of Insubria
6Department of Biomedical Sciences for Health, University of Milan
Abstract
Background: Overlapping systematic reviews (SRs) are increasingly frequent in the medical literature. They can easily originate discordant evidence. Reconciling conflicting evidence is a dimension not sufficiently addressed by guideline-development tools. As part of a wider research project supported by the Italian Ministry of Health, we are carrying out a survey to learn more about discordant SRs and their impact on the development of clinical recommendations (CRs).

Objectives: To identify key dimensions and informational needs that could be useful in dealing with discordant evidence in the context of the guideline decision-making process.

Methods: The survey consists of two parts. In the first one general information about the respondents is collected (ie. age, role, expertise, etc…). In the second one we present 4 GRADE Summary of Findings (SoF) tables summarising the results of 4 overlapping discordant SRs and 10 questions investigating the use of the evidence presented to take a decision about a possible CR. Moving from a real scenario, we explore the information needs when dealing with potential discordant evidence.

Results: The survey was sent to 80 people involved at different levels in the development of CRs. To date it has been completed by the 40% of the contacted people. First results show that the most-wanted information when in the presence of overlapping SRs are (multiple choice possible): Risk of Bias of SRs (69,2%); consistency between studies' results (65,4%); included studies in each SR (61,5%); and, methodological limitations in primary studies (50%).

Any response rate above 50% would be considered sufficient for a descriptive study. We will send a maximum of 3 reminders in a 2-month period.

Conclusions: The answers to the survey will help in identifying key dimensions and information needed when in presence of overlapping, discordant SRs. This will be used to develop a new module (Discordant Module) of the GRADE Working Group Guideline Development Tool (GDT - http://gdt.guidelinedevelopment.org) designed to support guideline developers in dealing with overlapping and discordant evidence.