We study how social proximity between the sender and the receiver of information shapes the effectiveness of preventive health campaigns. Focusing on shared religious affiliation as a signal of proximity, we implemented a field experiment during the initial phase of the COVID-19 pandemic in two major Indian cities characterized by Hindu-Muslim tensions. We randomly allocated a representative sample of slum residents to receive either “doctor messages” consisting of voice messages promoting recommended practices to prevent virus transmission or uninformative control messages on their mobile phones. The messages, introduced by a local citizen (the sender), were cross-randomized to commence with a greeting signaling either a Hindu or a Muslim identity, thereby manipulating the religion concordance between the sender and receiver. As compared to control messages, doctor messages are effective at improving compliance with recommended practices. However, the impact of these messages is contingent on the presence of religion concordance. In instances of concordance, the take-up of doctor messages is larger, and the intervention demonstrates efficacy in establishing a protective barrier against misinformation related to preventive practices.