BackgroundAs the rate of COVID-19 infections grew in Kisoro, Uganda, fear and misinformation about the virus were rife. Accurate, trustworthy community education seemed essential to support prevention efforts in the villages, allay widespread fear of death, and avoid the overwhelming of Kisoro District Hospital (KDH). Since 2005, KDH has collaborated with an NGO, Doctors for Global Health (DGH) and the Albert Einstein College of Medicine, NY, USA, to sponsor a robust Village Health Worker (VHW) programme in 52 villages in the Kisoro district. Community health education has been a cornerstone of the programme since its inception, and VHW-delivered home talks with portable, pictorial flip charts have shown success as a model for health education. Here, we describe a COVID-19 home-talk programme developed in a short time in response the COVID-19 pandemic and evaluate learning from this programme compared with learning from local radio (the main channel of information) and other regional information sources. In a COVID-19 lockdown, would the home-visit model be applicable? Would a health-worker delivered home-talk programme add to learning otherwise garnered from radio, television or neighbours? MethodsWe developed a 30 min COVID-19 home talk in 10 days; we trained 48 VHWs in an intense 2-day training, then monitored and certified VHW's skills over three sessions of field observation. Home talks were then fully implemented with a maximum of four adults per talk and social distancing was observed. To measure the retained learning from home talks, one adult per talk answered a six-item pre-test, and 3–5 weeks later, we randomly selected 20% of these participants and invited them to complete an identical post-test. To control for media exposure and assess contamination of the talk messages over time in study villages, residents of non-participating villages also completed tests at the same time that study participants completed post-tests in participating villages. FindingsIn the 3 months between April 20, 2020, and June 16, 2020, 48 VHWs gave 4308 COVID-19 home talks to more than 14 000 adults who have minimal education. Participants' post-test scores were 20% higher than pre-test scores and 30% higher than controls' test scores (p<0·0001) and were independent of the VHW's presentation skills. Significant learning was seen surrounding COVID-19 symptoms, mechanisms of spread, disease prevention, and risks of mortality, but not about when to go to the hospital with symptoms. Most participants (82%) reported understanding and valuing information from the COVID-19 home talks more than they did information heard via the radio. InterpretationDespite ubiquitous coverage of COVID-19 issues in the popular media, small group home talks by VHWs were preferred by rural villagers in Uganda and resulted in new, retained learning. We believe that this preference was probably the result of trust in the VHW and the opportunity to focus and ask questions. Corollary learning by word of mouth also occurred in home talk villages. FundingAlbert Einstein College of Medicine, Global Health.