Read on Twitter
1/18: "Results and lessons from a 2-round Delphi method experiment on the 1-dose/2-dose COVID vaccine question"
2/18: How can groups of experts or simply concerned citizens better arrive at consensus positions on important questions? How can we pool our knowledge effectively, particularly in rapidly evolving situations like the current pandemic?
3/18: In the 1950's, the RAND Corporation ( @RANDCorporation) developed a technique called the Delphi Method to improve technological forecasting. The technique is quite simple: a group of participants are given a prompt to which they respond anonymously.
4/18: The responses are distributed to the group (also anonymously) after which another round of responses are solicited. The aim is to encourage consensus building by enabling participants to consider the opinions of the other responders in an off-line, anonymous format.
5/18: As a warmup experiment in bottom-up consensus building and grassroots scientific activism, I organized a group of 5 physicians, scientists, and decision analysis experts to participate in a 2-round Delphi experiment to examine the COVID vaccine 1-dose/2-dose question.
6/18: Participants: Victor Faundez, MD, PhD, Professor of Cell Biology, Emory University School of Medicine ( @FaundezLab); Mike Rosen, PhD, Head of Computational Biology, D2Oncology;
7/18: Sam Chorlton, MD, Resident Physician in Medical Microbiology, University of British Columbia; David Manheim, PhD, Lead Researcher, 1 Day Sooner ( @davidmanheim); Gopal Sarma, MD, PhD, ML Strategy and Operations Lead, Broad Institute of MIT and Harvard ( @gopalpsarma)
8/18: The experiment was conducted using Google Forms and required very little time -- start to finish, it lasted 48 hours. Prompt: "Should countries consider a single-dose strategy for their COVID vaccine rollout in order to increase the speed of vaccine coverage?"
9/18: After the first round 3/5 were in favor of the 1-dose strategy and 2/5 were in favor of adhering to the regimen that was tested in the Pfizer/Moderna RCTs.
10/18: However, after the second round, all 5 participants were in agreement about trialing a 1-dose strategy, with the expectation of ongoing data collection and a potential reversal.
11/18: This isn't meant to be some profound revelation about policy. None of us who participated are experts in immunology or vaccine science and we have no intention of overstepping our boundaries.
12/18: The real purpose of this exercise was to understand the approach and to experiment with building consensus in a systematic way. It was also an opportunity to practice changing one's mind in a structured setting.
13/18: I personally found this to be a valuable exercise and I have also tested the Delphi method with product development questions with my team @broadinstitute
14/18: I'm writing this, because I want to challenge those experts and decision makers who *are* in a position of influencing policy during the pandemic to consider using the Delphi method among themselves, for vaccine dosing or other crucial decisions.
15/18: @EricTopol @skathire @Bob_Wachter @ashishkjha @VirusesImmunity. You have all had amazing careers and given others an incredible opportunity to learn from your wisdom and experience with your high quality Tweets during the pandemic.
16/18: You have also publicly expressed divergent viewpoints on the 1-dose/2-dose question. Would you consider running a Delphi experiment among yourselves?
17/18: The major issue I can foresee is that all of you are likely well aware of each other's viewpoints, so it's possible that there might not be much room for shifting beliefs at this point.
18/18: If that is the case, would you consider proposing this to your colleagues? Perhaps those serving on the federal COVID task force?
Fin.
Fin.
