Process Mining techniques have been successfully applied as a data-driven and domain-aware approach for improving business process performance in several organizations. Among its applications, Deviance Mining aims at uncovering the reasons why a subset of the executions of a business process deviate with respect to its expected or desirable outcomes, thus producing insights towards improving the process operation. Despite the fact that real-life processes are typically composed by non-atomic events, existing approaches for deviance mining only deal with atomic events in their experiments. This work proposes a domain-driven method for codifying processes with non-atomic events that uses the temporal dimension for discovering the most effective treatments that represent deviant process executions. The method is applied in the COVID-19 domain, to find who the temporal dimension of the non-pharmaceutical interventions contributed to slow down the rate of COVID-19 cases in countries around the world.