How we changed physician vaccine conversations

Duke University · Center for Advanced Hindsight · GSK partnership · 2023 – 2024

partnership Project manager Coding Publication Health behaviors

59.5% increase in presumptive language use Physicians trained to recommend vaccines more effectively — sustained at 90 days in two countries

Publication

Transforming provider–patient vaccine conversations: a randomised trial on presumptive communication training
Bartmann, N., Lindemans, J.W., Sloan, C.E., Vojtek, I., & van Wouw, M.
BMJ Leader, 2025 · Read · Open data (OSF)

Context

Despite widespread vaccine availability and favorable provider attitudes, immunization rates remain suboptimal worldwide. A healthcare provider’s recommendation is the single most effective driver of vaccine uptake — yet physicians don’t always recommend, and when they do, the language often isn’t persuasive enough.

Problem

GSK funded a K research program at Duke to close the gap between provider intentions and actual recommendation behavior. The question: can we train physicians to use presumptive (opt-out) language — “You’re due for your flu shot, we’ll do that at the end of this visit” instead of “Would you like the flu shot?” — and does the effect last?

Insight

We combined two behavioral levers: presumptive communication training (the AIMS method — Announce, Inquire, Mirror, Secure) and social norms messaging (showing providers that most peers already recommend vaccines). The hypothesis was that training alone would change language, while social norms would reinforce confidence and trust.

What I did

Study design: Two preregistered, longitudinal RCTs in the USA and Brazil — 413 physicians, 2×2 factorial (communication training × social norms), with follow-ups at 30, 60, and 90 days
Intervention: Built an interactive online training module (Qualtrics) teaching the AIMS method for vaccine conversations
Analysis: Linear mixed-effects models in R; data and code published on OSF
Project management: Coordinated cross-cultural execution across Duke, GSK, and a Brazilian recruitment partner; managed IRB, budgets, and timelines

Results

Physicians who received the training reported a 59.5% increase in presumptive language use in the USA (24.1 pp) and a 25.3% increase in Brazil (16.2 pp) at 90 days post-training. The effect was sustained across all follow-up time points. Social norms messaging increased perceived patient trust in Brazil at 60 and 90 days. Crucially, the training did not erode patient trust in either country.

← Home