Drafting evidence-based recommendations in 10 days in the context of the COVID-19 pandemic: methods and pitfalls

Article type
Authors
Franco JVA1, Escobar Liquitay CM1, Peuchot V2, Boietti B2, Sanchez Thomas D2, Torres Gomez F2, Peroni L2, Baez GN3, Pisula P3, Pollán J2
1Centro Cochrane, Instituto Universitario Hospital Italiano
2Servicio de Clínica Médica, Hospital Italiano de Buenos Aires
3Carrera de Medicina, Instituto Universitario Hospital Italiano
Abstract
Introduction
The arrival of the COVID-19 pandemic to our country introduced uncertainties regarding treatment options. Current recommendations were based on supportive care and the use of drugs capable of inhibiting viral replication such as hydroxychloroquine or lopinavir/ritonavir was suggested by other countries that faced the initial outbreaks of the pandemic.
Objectives
To generate rapid recommendations on available pharmacological interventions for the treatment COVID-19 and to identify the opportunities and pitfalls of this process in the context of ongoing pandemic disease.
Methods
We generated pragmatic searches in multiple databases searching direct and indirect evidence on the pharmacological treatments for COVID-19 (we excluded case reports, editorials and reviews). We created Summary of Findings tables and presented them to a multidisciplinary panel, including patients, to develop recommendations using explicit criteria under the Evidence to Decision Framework. The decision and the process were documented using a qualitative approach (notes and transcriptions of the working sessions).
Results
We screened 3191 references (March 21st 2020) and included 1 RCT for ritonavir/lopinavir and 1 observational study for hydroxychloroquine. Based on low to very-low quality of evidence these interventions caused little to no difference in the pre-defined critical outcomes identified by our community. Ten days after starting the project we had a virtual meeting with the panel and a decision was made to recommend the use of lopinavir/ritonavir for critically ill adults (weak recommendation) and if the patient had a contraindication to lopinavir/ritonavir, then use hydroxychloroquine (weak recommendation). Out main difficulties were: the emerging available evidence from primary studies, usually in forms of preprints, or available Summary of Findings tables from rapid reviews or health technology assessments (many of them with different outcomes, data or interpretation). Furthermore, scientific societies and local and foreign governments issued simultaneously conflicting recommendations, which caused some reluctance to the development of the first evidence-based recommendations of our institution.
Conclusions
We were able to generate rapid recommendations for COVID-19 in a short time span (10 days), however, we faced emerging methodological challenges in the evidence synthesis process and contextual challenges when formulating recommendations. This experience could help in the development of methodological guidance for recommendations in the context of an emergency such as pandemic disease.
Consumer involvement:
We conducted an open online poll to people in the community and healthcare professionals (>400 responses) to identify critical outcomes for SoF tables. A patient advocate was incorporated into the decision panel.
Patient or healthcare consumer involvement: